openapi: 3.1.0
info:
  title: Fireworks Text Completion API
  version: 0.1.0
  description: Text completion API for LLM inference
paths:
  /v1/completions:
    post:
      summary: Create Completion
      description: "Create a completion for the provided prompt and parameters.


        For RL / agent rollouts, Fireworks inference exposes additional

        rollout-specific features:

        [`x-session-affinity` and `x-multi-turn-session-id`](https://docs.fireworks.ai/guides/rollout-inference#session-affinity)

        for multi-turn trajectories, and

        [MoE Router Replay (R3)](https://docs.fireworks.ai/guides/rollout-inference#moe-router-replay)

        for MoE expert tracing during rollouts."
      operationId: create_completion_v1_completions_post
      requestBody:
        content:
          application/json:
            schema:
              $ref: "#/components/schemas/CompletionRequest"
        required: true
      responses:
        "200":
          description: Successful Response
          content:
            application/json:
              schema:
                $ref: "#/components/schemas/CompletionResponse"
            text/event-stream:
              schema:
                $ref: "#/components/schemas/CompletionStreamResponse"
        "422":
          description: Validation Error
          content:
            application/json:
              schema:
                $ref: "#/components/schemas/HTTPValidationError"
  /v1/chat/completions:
    post:
      summary: Create Chat Completion
      description: "Create a completion for the provided prompt and parameters.


        For RL / agent rollouts, Fireworks inference exposes additional

        rollout-specific features:

        [`x-session-affinity` and `x-multi-turn-session-id`](https://docs.fireworks.ai/guides/rollout-inference#session-affinity)

        for multi-turn trajectories, and

        [MoE Router Replay (R3)](https://docs.fireworks.ai/guides/rollout-inference#moe-router-replay)

        for MoE expert tracing during rollouts."
      operationId: create_chat_completion_v1_chat_completions_post
      requestBody:
        content:
          application/json:
            schema:
              $ref: "#/components/schemas/ChatCompletionRequest"
        required: true
      responses:
        "200":
          description: Successful Response
          content:
            application/json:
              schema:
                $ref: "#/components/schemas/ChatCompletionResponse"
            text/event-stream:
              schema:
                $ref: "#/components/schemas/ChatCompletionStreamResponse"
        "422":
          description: Validation Error
          content:
            application/json:
              schema:
                $ref: "#/components/schemas/HTTPValidationError"
components:
  schemas:
    AllowedToolsConfig:
      properties:
        mode:
          type: string
          enum:
            - auto
            - required
          title: Mode
          description:
            "`auto` lets the model pick between generating a message and
            calling one of the allowed tools. `required` forces the model to call
            one of them."
        tools:
          items:
            additionalProperties: true
            type: object
          type: array
          title: Tools
          description:
            'A list of tool references the model is allowed to invoke.
            Each entry must be of the form `{"type": "function", "function": {"name":
            "..."}}`. The referenced tools must also appear in the top-level `tools`
            list.'
      additionalProperties: false
      type: object
      required:
        - mode
        - tools
      title: AllowedToolsConfig
      description: 'OpenAI-compatible allowed_tools sub-object.


        See https://platform.openai.com/docs/guides/function-calling#tool-choice


        The ``tools`` list mirrors the ``tools`` list passed at the top level: each

        entry must be ``{"type": "function", "function": {"name": "<fn>"}}``.

        Fireworks only supports function tools today, so other ``type`` values are

        rejected.'
    AllowedToolsSelection:
      properties:
        type:
          type: string
          const: allowed_tools
          title: Type
        allowed_tools:
          $ref: "#/components/schemas/AllowedToolsConfig"
      additionalProperties: false
      type: object
      required:
        - type
        - allowed_tools
      title: AllowedToolsSelection
      description:
        'OpenAI-compatible ``tool_choice.type = "allowed_tools"`` selection.


        Restricts the model''s tool calls to a subset of the top-level ``tools`` list

        *without modifying the prompt*, preserving prompt cache hits. The full set

        of tool definitions is still rendered into the prompt for the model; only

        the grammar (and downstream parsing) is constrained to the allowed subset.'
    ChatCompletionFunction:
      properties:
        name:
          type: string
          title: Name
          description:
            The name of the function to be called. Must be a-z, A-Z, 0-9,
            or contain underscores and dashes, with a maximum length of 64.
        description:
          anyOf:
            - type: string
            - type: "null"
          title: Description
          description:
            A description of what the function does, used by the model
            to choose when and how to call the function.
        parameters:
          additionalProperties: true
          type: object
          title: Parameters
          description:
            "The parameters the function accepts, described as a JSON Schema\
            \ object.\n\nThe JSON Schema object should have the following structure:\n\
            \n```json\n{\n  \"type\": \"object\",\n  \"required\": [\"param1\", \"\
            param2\"],\n  \"properties\": {\n    \"param1\": {\n      \"type\": \"\
            string\",\n      \"description\": \"...\"\n    },\n    \"param2\": {\n\
            \      \"type\": \"number\",\n      \"description\": \"...\"\n    }\n\
            \  }\n}\n```\n\n- The `type` field must be `\"object\"`.\n- The `required`\
            \ field is an array of strings indicating which parameters are required.\n\
            - The `properties` field is a map of property names to their definitions,\
            \ where each property is an object with `type` (string) and `description`\
            \ (string) fields.\n\nTo describe a function that accepts no parameters,\
            \ provide the value:\n\n```json\n{\"type\": \"object\", \"properties\"\
            : {}}\n```"
        strict:
          anyOf:
            - type: boolean
            - type: "null"
          title: Strict
      additionalProperties: false
      type: object
      required:
        - name
      title: ChatCompletionFunction
    ChatCompletionMessageToolCall:
      properties:
        id:
          anyOf:
            - type: string
            - type: "null"
          title: Id
          description: The ID of the tool call.
        type:
          type: string
          title: Type
          description: The type of the tool. Currently, only `function` is supported.
          default: function
        function:
          anyOf:
            - $ref: "#/components/schemas/ChatCompletionMessageToolCallFunction"
            - type: string
          title: Function
          description: The function that the model called.
      additionalProperties: false
      type: object
      required:
        - function
      title: ChatCompletionMessageToolCall
    ChatCompletionMessageToolCallFunction:
      properties:
        name:
          anyOf:
            - type: string
            - type: "null"
          title: Name
          description: The name of the function to call.
        arguments:
          anyOf:
            - type: string
            - additionalProperties: true
              type: object
            - type: "null"
          title: Arguments
          description:
            The arguments to call the function with, as generated by the
            model in JSON format. Note that the model does not always generate valid
            JSON, and may hallucinate parameters not defined by your function schema.
            Validate the arguments in your code before calling your function.
      additionalProperties: false
      type: object
      title: ChatCompletionMessageToolCallFunction
    ChatCompletionRequest:
      properties:
        model:
          type: string
          title: Model
          description: 'The name of the model to use.


            Example: `"accounts/fireworks/models/kimi-k2-instruct-0905"`'
        messages:
          items:
            $ref: "#/components/schemas/ChatMessage"
          type: array
          title: Messages
          description:
            A list of messages comprising the conversation so far. When
            ``prompt_token_ids`` is supplied this field is ignored (the caller is
            taking responsibility for chat-template rendering and tokenization upstream).
        tools:
          items:
            $ref: "#/components/schemas/ChatCompletionTool"
          type: array
          title: Tools
          description:
            "A list of tools the model may call. Currently, only functions
            are supported as a tool.


            Use this to provide a list of functions the model may generate JSON inputs
            for.


            See the our [model library](https://app.fireworks.ai/models/?filter=LLM&functionCalling=true)
            for the list of supported models"
        tool_choice:
          anyOf:
            - type: string
              enum:
                - auto
                - none
                - any
                - required
            - $ref: "#/components/schemas/FunctionSelection"
            - $ref: "#/components/schemas/AllowedToolsSelection"
          title: Tool Choice
          description:
            "Controls which (if any) tool is called by the model.\n\n-\
            \ `none`: the model will not call any tool and instead generates a message.\n\
            - `auto`: the model can pick between generating a message or calling one\
            \ or more tools.\n- `required` (alias: `any`): the model must call one\
            \ or more tools.\n  To force a specific function, pass an object of the\
            \ form `{ \"type\": \"function\", \"name\": \"my_function\" }` or `{ \"\
            type\": \"function\", \"function\": { \"name\": \"my_function\" } }` for\
            \ OpenAI compatibility.\n- To restrict tool calls to a subset of the provided\
            \ tools while keeping the\n  full tool list in the prompt (preserving\
            \ prompt cache hits), pass\n  `{ \"type\": \"allowed_tools\", \"allowed_tools\"\
            : { \"mode\": \"auto\"|\"required\", \"tools\": [{ \"type\": \"function\"\
            , \"function\": { \"name\": \"...\" } }] } }`.\n  Equivalent OpenAI Responses-style\
            \ flat form is also accepted: `{ \"type\": \"allowed_tools\", \"mode\"\
            : \"...\", \"tools\": [...] }`."
          default: auto
        stream:
          anyOf:
            - type: boolean
            - type: "null"
          title: Stream
          description:
            "Whether to stream back partial progress. If set, tokens will
            be sent as data-only [server-sent events](https://developer.mozilla.org/en-US/docs/Web/API/Server-sent_events/Using_server-sent_events#Event_stream_format)
            as they become available, with the stream terminated by a `data: [DONE]`
            message."
          default: false
        response_format:
          anyOf:
            - $ref: "#/components/schemas/ResponseFormat"
            - type: "null"
          description:
            'Allows to force the model to produce specific output format.


            Setting to `{ "type": "json_object" }` enables JSON mode, which guarantees
            the message the model generates is valid JSON.


            If `"type"` is `"json_schema"`, a JSON schema must be provided. E.g.,
            `response_format = {"type": "json_schema", "json_schema": <json_schema>}`.


            Important: when using JSON mode, it''s crucial to also instruct the model
            to produce JSON via a system or user message. Without this, the model
            may generate an unending stream of whitespace until the generation reaches
            the token limit, resulting in a long-running and seemingly "stuck" request.


            Also note that the message content may be partially cut off if `finish_reason="length"`,
            which indicates the generation exceeded `max_tokens` or the conversation
            exceeded the max context length. In this case the return value might not
            be a valid JSON.'
        temperature:
          anyOf:
            - type: number
            - type: "null"
          title: Temperature
          description:
            "What sampling temperature to use, between 0 and 2. Higher
            values like 0.8 will make the output more random, while lower values like
            0.2 will make it more focused and deterministic.


            We generally recommend altering this or top_p but not both.


            Required range: `0 <= x <= 2`


            Example: `1`"
        top_k:
          anyOf:
            - type: integer
            - type: "null"
          title: Top K
          description:
            "Top-k sampling is another sampling method where the k most
            probable next tokens are filtered and the probability mass is redistributed
            among only those k next tokens. The value of k controls the number of
            candidates for the next token at each step during text generation. Must
            be between 0 and 100.


            Required range: `0 <= x <= 100`


            Example: `50`"
        user:
          anyOf:
            - type: string
            - type: "null"
          title: User
          description:
            A unique identifier representing your end-user, which can help
            monitor and detect abuse.
        prompt_cache_key:
          anyOf:
            - type: string
            - type: "null"
          title: Prompt Cache Key
          description:
            A key used for prompt caching session affinity. Requests with
            the same prompt_cache_key are routed to the same backend to maximize KV
            cache hit rates. This is the preferred field for session affinity (takes
            priority over the 'user' field).
        prompt_cache_isolation_key:
          anyOf:
            - type: string
            - type: "null"
          title: Prompt Cache Isolation Key
          description: Isolation key for prompt caching to separate cache entries.
        raw_output:
          anyOf:
            - type: boolean
            - type: "null"
          title: Raw Output
          description: Return raw output from the model.
          default: false
        perf_metrics_in_response:
          anyOf:
            - type: boolean
            - type: "null"
          title: Perf Metrics In Response
          description:
            "Whether to include performance metrics in the response body.


            **Non-streaming requests:** Performance metrics are always included in
            response headers (e.g., `fireworks-prompt-tokens`, `fireworks-server-time-to-first-token`).
            Setting this to `true` additionally includes the same metrics in the response
            body under the `perf_metrics` field.


            **Streaming requests:** Performance metrics are only included in the response
            body under the `perf_metrics` field in the final chunk (when `finish_reason`
            is set). This is because headers may not be accessible during streaming.


            The response body `perf_metrics` field contains the following metrics:


            **Basic Metrics (all deployments):**


            - `prompt-tokens`: Number of tokens in the prompt

            - `cached-prompt-tokens`: Number of cached prompt tokens

            - `server-time-to-first-token`: Time from request start to first token
            (in seconds)

            - `server-processing-time`: Total processing time (in seconds, only for
            completed requests)


            **Predicted Outputs Metrics:**


            - `speculation-prompt-tokens`: Number of speculative prompt tokens

            - `speculation-prompt-matched-tokens`: Number of matched speculative prompt
            tokens (for completed requests)


            **Dedicated Deployment Only Metrics:**


            - `speculation-generated-tokens`: Number of speculative generated tokens
            (for completed requests)

            - `speculation-acceptance`: Speculation acceptance rates by position

            - `backend-host`: Hostname of the backend server

            - `num-concurrent-requests`: Number of concurrent requests

            - `deployment`: Deployment name

            - `tokenizer-queue-duration`: Time spent in tokenizer queue

            - `tokenizer-duration`: Time spent in tokenizer

            - `prefill-queue-duration`: Time spent in prefill queue

            - `prefill-duration`: Time spent in prefill

            - `generation-queue-duration`: Time spent in generation queue

            - `generation-duration`: Time spent in generation"
          default: false
        stream_options:
          anyOf:
            - $ref: "#/components/schemas/StreamOptions"
            - type: "null"
          description:
            Options for streaming responses. Only valid with ``stream=true``.
            Fireworks includes a final SSE chunk carrying usage totals by default;
            set ``include_usage=false`` to opt out. The ``buffer_tokens`` / ``buffer_ms``
            / ``buffer_mode`` fields coalesce SSE chunks by token count and/or time
            (vLLM backend; override the deployment default).
        n:
          type: integer
          title: N
          description: "How many completions to generate for each prompt.


            **Note:** Because this parameter generates many completions, it can quickly
            consume your token quota. Use carefully and ensure that you have reasonable
            settings for `max_tokens` and `stop`.


            Required range: `1 <= x <= 128`


            Example: `1`"
          default: 1
        service_tier:
          type: string
          enum:
            - auto
            - default
            - flex
            - priority
          title: Service Tier
          description:
            The service tier to use for the request. Specifies the processing
            type used for serving the request. Only "priority" is supported, while
            all other values will be treated as "default" tier.
          default: default
        stop:
          anyOf:
            - type: string
            - items:
                type: string
              type: array
            - type: "null"
          title: Stop
          description:
            Up to 4 sequences where the API will stop generating further
            tokens. The returned text will NOT contain the stop sequence.
        max_tokens:
          anyOf:
            - type: integer
            - type: "null"
          title: Max Tokens
          description:
            The maximum number of tokens to generate in the completion.
            If the token count of your prompt plus max_tokens exceeds the model's
            context length, the behavior depends on context_length_exceeded_behavior.
            By default, max_tokens will be lowered to fit in the context window instead
            of returning an error.
        max_completion_tokens:
          anyOf:
            - type: integer
            - type: "null"
          title: Max Completion Tokens
          description: Alias for max_tokens. Cannot be specified together with max_tokens.
        top_p:
          anyOf:
            - type: number
            - type: "null"
          title: Top P
          description:
            "An alternative to sampling with temperature, called nucleus
            sampling, where the model considers the results of the tokens with top_p
            probability mass. So 0.1 means only the tokens comprising the top 10%
            probability mass are considered.


            We generally recommend altering this or temperature but not both.


            Required range: `0 <= x <= 1`


            Example: `1`"
        min_p:
          anyOf:
            - type: number
            - type: "null"
          title: Min P
          description:
            "Minimum probability threshold for token selection. Only tokens
            with probability >= min_p are considered for selection. This is an alternative
            to `top_p` and `top_k` sampling.


            Required range: `0 <= x <= 1`"
        typical_p:
          anyOf:
            - type: number
            - type: "null"
          title: Typical P
          description:
            "Typical-p sampling is an alternative to nucleus sampling.
            It considers the most typical tokens whose cumulative probability is at
            most typical_p.


            Required range: `0 <= x <= 1`"
        frequency_penalty:
          anyOf:
            - type: number
            - type: "null"
          title: Frequency Penalty
          description:
            "Number between -2.0 and 2.0. Positive values penalize new
            tokens based on their existing frequency in the text so far, decreasing
            the model's likelihood to repeat the same line verbatim.


            Reasonable value is around 0.1 to 1 if the aim is to just reduce repetitive
            samples somewhat. If the aim is to strongly suppress repetition, then
            one can increase the coefficients up to 2, but this can noticeably degrade
            the quality of samples. Negative values can be used to increase the likelihood
            of repetition.


            See also `presence_penalty` for penalizing tokens that have at least one
            appearance at a fixed rate.


            OpenAI compatible (follows OpenAI's conventions for handling token frequency
            and repetition penalties).


            Required range: `-2 <= x <= 2`"
        presence_penalty:
          anyOf:
            - type: number
            - type: "null"
          title: Presence Penalty
          description:
            "Number between -2.0 and 2.0. Positive values penalize new
            tokens based on whether they appear in the text so far, increasing the
            model's likelihood to talk about new topics.


            Reasonable value is around 0.1 to 1 if the aim is to just reduce repetitive
            samples somewhat. If the aim is to strongly suppress repetition, then
            one can increase the coefficients up to 2, but this can noticeably degrade
            the quality of samples. Negative values can be used to increase the likelihood
            of repetition.


            See also `frequency_penalty` for penalizing tokens at an increasing rate
            depending on how often they appear.


            OpenAI compatible (follows OpenAI's conventions for handling token frequency
            and repetition penalties).


            Required range: `-2 <= x <= 2`"
        repetition_penalty:
          anyOf:
            - type: number
            - type: "null"
          title: Repetition Penalty
          description:
            "Applies a penalty to repeated tokens to discourage or encourage
            repetition. A value of `1.0` means no penalty, allowing free repetition.
            Values above `1.0` penalize repetition, reducing the likelihood of repeating
            tokens. Values between `0.0` and `1.0` reward repetition, increasing the
            chance of repeated tokens. For a good balance, a value of `1.2` is often
            recommended. Note that the penalty is applied to both the generated output
            and the prompt in decoder-only models.


            Required range: `0 <= x <= 2`"
        mirostat_target:
          anyOf:
            - type: number
            - type: "null"
          title: Mirostat Target
          description:
            "Defines the target perplexity for the Mirostat algorithm.
            Perplexity measures the unpredictability of the generated text, with higher
            values encouraging more diverse and creative outputs, while lower values
            prioritize predictability and coherence. The algorithm dynamically adjusts
            the token selection to maintain this target during text generation.


            If not specified, Mirostat sampling is disabled."
        mirostat_lr:
          anyOf:
            - type: number
            - type: "null"
          title: Mirostat Lr
          description:
            Specifies the learning rate for the Mirostat sampling algorithm,
            which controls how quickly the model adjusts its token distribution to
            maintain the target perplexity. A smaller value slows down the adjustments,
            leading to more stable but gradual shifts, while higher values speed up
            corrections at the cost of potential instability.
        seed:
          anyOf:
            - type: integer
            - type: "null"
          title: Seed
          description: Random seed for deterministic sampling.
        logprobs:
          anyOf:
            - type: integer
            - type: boolean
            - type: "null"
          title: Logprobs
          description:
            "Include log probabilities in the response. This accepts either
            a boolean or an integer:


            If set to `true`, log probabilities are included and the number of alternatives
            can be controlled via `top_logprobs` (OpenAI-compatible behavior).


            If set to an integer N, include log probabilities for up to N most likely
            tokens per position in the legacy format. N must be between 0 and the
            deployment's `--max-logprobs` limit (5 by default).


            The API will always return the logprob of the sampled token, so there
            may be up to `logprobs+1` elements in the response when an integer is
            used."
        top_logprobs:
          anyOf:
            - type: integer
            - type: "null"
          title: Top Logprobs
          description:
            "An integer specifying the number of most likely tokens to
            return at each token position, each with an associated log probability.
            Must be between 0 and the deployment's `--max-logprobs` limit (5 by default).


            When `logprobs` is set, `top_logprobs` can be used to modify how many
            top log probabilities are returned. If `top_logprobs` is not set, the
            API will return up to `logprobs` tokens per position."
        sampling_mask:
          anyOf:
            - type: string
              enum:
                - count
                - non_zero_list
                - non_zero_buffer
            - type: "null"
          title: Sampling Mask
          description:
            Opt-in sampling mask metadata for generated tokens. When set
            to `"count"`, each generated token in the new logprobs format includes
            the number of token logits still eligible for sampling after filters such
            as top_p and top_k are applied. `"non_zero_list"` additionally returns
            active token IDs in `sampling_mask`; `"non_zero_buffer"` additionally
            returns a base64-encoded little-endian uint32 buffer of active token IDs.
            Non-zero payloads are omitted for positions with more active tokens than
            1000.
        echo:
          anyOf:
            - type: boolean
            - type: "null"
          title: Echo
          description: Echo back the prompt in addition to the completion.
          default: false
        echo_last:
          anyOf:
            - type: integer
            - type: "null"
          title: Echo Last
          description:
            Echo back the last N tokens of the prompt in addition to the
            completion. This is useful for obtaining logprobs of the prompt suffix
            but without transferring too much data. Passing `echo_last=len(prompt)`
            is the same as `echo=True`
        ignore_eos:
          type: boolean
          title: Ignore Eos
          description:
            This setting controls whether the model should ignore the End
            of Sequence (EOS) token. When set to `True`, the model will continue generating
            tokens even after the EOS token is produced. By default, it stops when
            the EOS token is reached.
          default: false
        context_length_exceeded_behavior:
          type: string
          enum:
            - error
            - truncate
          title: Context Length Exceeded Behavior
          description:
            "What to do if the token count of prompt plus `max_tokens`
            exceeds the model's context window.


            Passing `truncate` limits the `max_tokens` to at most `context_window_length
            - prompt_length`. This is the default.


            Passing `error` would trigger a request error.


            The default of `'truncate'` is selected as it allows to ask for high
            `max_tokens` value while respecting the context window length without
            having to do client-side prompt tokenization.


            Note, that it differs from OpenAI's behavior that matches that of `error`."
          default: truncate
        logit_bias:
          anyOf:
            - additionalProperties:
                type: number
              type: object
            - type: "null"
          title: Logit Bias
          description:
            Modify the likelihood of specified tokens appearing in the
            completion. Accepts a json object that maps tokens (specified by their
            token ID in the tokenizer) to an associated bias value from -100 to 100.
            Mathematically, the bias is added to the logits generated by the model
            prior to sampling.
        speculation:
          anyOf:
            - type: string
            - items:
                type: integer
              type: array
            - type: "null"
          title: Speculation
          description: Speculative decoding prompt or token IDs to speed up generation.
        prediction:
          anyOf:
            - $ref: "#/components/schemas/PredictedOutput"
            - type: string
            - type: "null"
          title: Prediction
          description:
            OpenAI-compatible predicted output for speculative decoding.
            Can be a PredictedOutput object or a simple string. Automatically transformed
            to speculation.
        metadata:
          anyOf:
            - additionalProperties:
                type: string
              type: object
            - type: "null"
          title: Metadata
          description: Additional metadata to store with the request for tracing/distillation.
        reasoning_effort:
          anyOf:
            - type: string
              enum:
                - low
                - medium
                - high
                - xhigh
                - max
                - none
                - adaptive
            - type: integer
            - type: boolean
            - type: "null"
          title: Reasoning Effort
          description:
            'Controls reasoning behavior for supported models. When enabled,
            the model''s reasoning appears in the `reasoning_content` field of the
            response, separate from the final answer in `content`.


            **Accepted values:**


            - **String** (OpenAI-compatible): `''low''`, `''medium''`, `''high''`,
            or `''max''` to enable reasoning with varying effort levels; `''none''`
            to disable reasoning.

            - **Boolean** (Fireworks extension): `true` to enable reasoning, `false`
            to disable it.

            - **Integer** (Fireworks extension): A positive integer to set a hard
            token limit on reasoning output. Integer values enable the model''s normal
            medium-style thinking behavior and force the model to end its thinking
            phase after at most that many generated thinking tokens.


            **Important:** Boolean values are normalized internally: `true` becomes
            `''medium''`, and `false` becomes `''none''`. This normalization happens
            before model-specific validation, so if a model doesn''t support `''none''`,
            passing `false` will produce an error referencing `''none''`.


            **Model-specific behavior:**


            - **Qwen3**: Grammar-based reasoning on the reasoning-enabled `qwen3`/`qwen3p5`
            conversation styles. Older chat-mode Qwen3 deployments may opt into `qwen3-no-thinking`,
            which disables reasoning support. For reasoning-enabled styles, use `''none''`
            or `false` to disable. Supports integer token limits to cap reasoning
            output. `''low''`, `''medium''`, and `''high''` keep their model-specific
            behavior and are not hard budgets.

            - **MiniMax M2**: Reasoning is required (always on). Defaults to `''medium''`
            when omitted. Accepts only string `reasoning_effort`: `''low''`, `''medium''`,
            or `''high''`. `''none''` and boolean values are rejected.

            - **DeepSeek V3.1**: Binary on/off reasoning. Default reasoning off (matches
            chat template). Use `true`, `''low''`, `''medium''`, or `''high''` to
            enable; `''none''` or `false` to disable.

            - **DeepSeek V3.2**: Binary on/off reasoning. Default reasoning on. Use
            `''none''` or `false` to disable; effort levels and integers have no additional
            effect.

            - **DeepSeek V4**: Accepts `''none''`, `''low''`, `''medium''`, `''high''`,
            `''xhigh''`, and `''max''`. Default reasoning on (`''high''`). `''xhigh''`
            is silently promoted to `''max''`. `''max''` prepends a thorough-reasoning
            preamble; `''high''` enables thinking. `''low''` and `''medium''` are
            silently promoted to `''high''`. `''none''` or `false` disables thinking.

            - **GLM 4.5, GLM 4.5 Air, GLM 4.6, GLM 4.7, GLM 5.1**: Binary on/off reasoning.
            Default reasoning on. Use `''none''` or `false` to disable; effort levels
            and integers have no additional effect.

            - **GLM 5.2**: Two thinking tiers, `High` and `Max` (rendered as a `Reasoning
            Effort:` system line). `''high''` selects High; `''low''` and `''medium''`
            are collapsed to `''high''`; `''max''` and `''xhigh''` select Max; when
            omitted, the model default (`Max`) applies. `''none''` or `false` disables
            thinking.

            - **Harmony (OpenAI GPT-OSS 120B, GPT-OSS 20B)**: Accepts only `''low''`,
            `''medium''`, or `''high''`. Does not support `''none''`, `false`, or
            integer values — using these will return an error (e.g., "Invalid reasoning
            effort: none"). When omitted, defaults to `''medium''`. Lower effort produces
            faster responses with shorter reasoning.'
        reasoning_history:
          anyOf:
            - type: string
              enum:
                - disabled
                - interleaved
                - preserved
            - type: "null"
          title: Reasoning History
          description:
            "Controls how historical assistant reasoning content is included
            in the prompt for multi-turn conversations.


            **Accepted values:**


            - `null`: Use model/template default behavior (for **GLM-4.7**, the model/template
            default is `'interleaved'`, i.e. historical reasoning is cleared by
            default)

            - `'disabled'`: Strip `reasoning_content` from all messages before prompt
            construction

            - `'interleaved'`: Strip `reasoning_content` from messages up to (and
            including) the last user message

            - `'preserved'`: Preserve historical `reasoning_content` across the
            conversation


            **Model support:**


            | Model | Default | Supported values |

            | --- | --- | --- |

            | Kimi K2.7 | `'preserved'` | `'disabled'`, `'interleaved'`, `'preserved'`
            |

            | Kimi K2.6 | `'interleaved'` | `'disabled'`, `'interleaved'`, `'preserved'`
            |

            | Kimi K2 Instruct | `'preserved'` | `'disabled'`, `'interleaved'`,
            `'preserved'` |

            | MiniMax M2 | `'interleaved'` | `'disabled'`, `'interleaved'` |

            | GLM-5.2 | `'interleaved'` | `'disabled'`, `'interleaved'`, `'preserved'`
            |

            | GLM-4.7 | `'interleaved'` | `'disabled'`, `'interleaved'`, `'preserved'`
            |

            | GLM-4.6 | `'interleaved'` | `'disabled'`, `'interleaved'` |

            | Qwen 3.6 | `'preserved'` | `'disabled'`, `'preserved'` |

            | DeepSeek V4 | `'interleaved'` | `'interleaved'` |


            For other models, refer to the model provider's documentation.


            **Note:** This parameter controls prompt formatting only. To disable reasoning
            computation entirely, use `reasoning_effort='none'`."
        thinking:
          anyOf:
            - $ref: "#/components/schemas/ThinkingConfigEnabled"
            - $ref: "#/components/schemas/ThinkingConfigDisabled"
            - $ref: "#/components/schemas/ThinkingConfigAdaptive"
            - type: "null"
          title: Thinking
          description:
            'Configuration for enabling extended thinking (Anthropic-compatible
            format). This is an alternative to `reasoning_effort` for controlling
            reasoning behavior.


            **Format:**


            - `{"type": "enabled"}` - Enable thinking (equivalent to `reasoning_effort:
            true`)

            - `{"type": "enabled", "budget_tokens": <int>}` - Enable thinking with
            a token budget (equivalent to `reasoning_effort: <int>`). Must be >= 1024.

            - `{"type": "enabled", "keep": "all"}` - Enable thinking and preserve
            all historical reasoning content in the prompt (equivalent to `reasoning_history:
            "preserved"`).

            - `{"type": "disabled"}` - Disable thinking (equivalent to `reasoning_effort:
            "none"`)


            **Precedence with `reasoning_effort`:** `thinking.effort` (when set) overrides
            `reasoning_effort`; otherwise, for `type=enabled`, `reasoning_effort`
            is used as the effort level. `type=disabled` always disables thinking.'
        return_token_ids:
          anyOf:
            - type: boolean
            - type: "null"
          title: Return Token Ids
          description: Return token IDs alongside text to avoid retokenization drift.
          default: false
        prompt_token_ids:
          anyOf:
            - items:
                type: integer
              type: array
            - type: "null"
          title: Prompt Token Ids
          description:
            "Pre-tokenized prompt. When set, the server skips chat-template
            rendering and tokenization for the input — the supplied token IDs are
            fed directly to generation. Generation, the response formatter (incl.
            ``--formatter-backend vllm`` parsers), and tool-call routing all run as
            usual on the model's output.


            Intended for upstream gateways (e.g. dynamo) that own the tokenizer and
            chat template, and want this server only to generate and format. Mutually
            exclusive with ``messages``: passing both is a 400."
        functions:
          items:
            $ref: "#/components/schemas/ChatCompletionFunction"
          type: array
          title: Functions
          description:
            Deprecated in OpenAI. Use 'tools' instead. This will be automatically
            transformed to tools.
          deprecated: true
        prompt_truncate_len:
          anyOf:
            - type: integer
            - type: "null"
          title: Prompt Truncate Len
          description:
            'The size (in tokens) to which to truncate chat prompts. This
            includes the system prompt (if any), previous user/assistant messages,
            and the current user message. Earlier user/assistant messages will be
            evicted first to fit the prompt into this length. The system prompt is
            preserved whenever possible and only truncated as a last resort.


            This should usually be set to a number much smaller << than the model''s
            maximum context size, to allow enough remaining tokens for generating
            a response.


            If omitted, you may receive "prompt too long" errors in your responses
            as conversations grow. Note that even with this set, you may still receive
            "prompt too long" errors if individual messages (such as a very long system
            prompt or user message) exceed the model''s context window on their own.'
        parallel_tool_calls:
          anyOf:
            - type: boolean
            - type: "null"
          title: Parallel Tool Calls
          description: Enable parallel function calling.
        safe_tokenization:
          anyOf:
            - type: boolean
            - type: "null"
          title: Safe Tokenization
          description:
            "When true, special tokens in user-provided content are never
            interpreted as actual special tokens during tokenization. This prevents
            prompt injection via special token strings (e.g. <|im_start|>, <｜User｜>).
            Supported for models using Jinja or HuggingFace chat templates with HuggingFace
            tokenizers. Explicit true returns an error if the model does not support
            it, or if combined with custom_chat_template on HuggingFace-backed models.
            Note: prompt_truncate_len is not applied when safe_tokenization is enabled.
            When omitted, safe tokenization is enabled by default on a best-effort
            basis: it is applied only where the model supports it and the request
            is text-only with no custom_chat_template and no prompt_truncate_len,
            and otherwise falls back to normal tokenization."
        function_call:
          anyOf:
            - type: string
              enum:
                - auto
                - none
            - $ref: "#/components/schemas/FunctionNameSpec"
            - type: "null"
          title: Function Call
          description:
            Deprecated in OpenAI. Use 'tool_choice' instead. This will
            be automatically transformed to tool_choice.
          deprecated: true
      additionalProperties: false
      type: object
      required:
        - model
      title: ChatCompletionRequest
    ChatCompletionTool:
      properties:
        type:
          type: string
          enum:
            - function
          title: Type
          description: The type of the tool. Currently, only `function` is supported.
        function:
          anyOf:
            - $ref: "#/components/schemas/ChatCompletionFunction"
            - type: "null"
          description: Required for function tools.
      additionalProperties: false
      type: object
      required:
        - type
      title: ChatCompletionTool
    ChatMessage:
      properties:
        role:
          type: string
          title: Role
          description:
            The role of the messages author. One of `system`, `user`, or
            `assistant`.
        content:
          anyOf:
            - type: string
            - items:
                $ref: "#/components/schemas/ChatMessageContent"
              type: array
            - type: "null"
          title: Content
          description:
            The contents of the message. `content` is required for all
            messages, and may be null for assistant messages with function calls.
        reasoning_content:
          anyOf:
            - type: string
            - type: "null"
          title: Reasoning Content
          description:
            The reasoning or thinking process generated by the model. This
            field is only available for certain reasoning models (GLM 4.5, GLM 4.5
            Air, GPT OSS 120B, GPT OSS 20B) and contains the model's internal reasoning
            that would otherwise appear in `<think></think>` tags within the content
            field.
        tool_calls:
          anyOf:
            - items:
                $ref: "#/components/schemas/ChatCompletionMessageToolCall"
              type: array
            - type: "null"
          title: Tool Calls
          description: The tool calls generated by the model, such as function calls.
        tool_call_id:
          anyOf:
            - type: string
            - type: "null"
          title: Tool Call Id
      additionalProperties: false
      type: object
      required:
        - role
      title: ChatMessage
      description: A chat completion message.
    ChatMessageContent:
      properties:
        type:
          type: string
          title: Type
        text:
          anyOf:
            - type: string
            - type: "null"
          title: Text
        image_url:
          anyOf:
            - $ref: "#/components/schemas/ChatMessageContentImageURL"
            - type: "null"
        video_url:
          anyOf:
            - $ref: "#/components/schemas/ChatMessageContentVideoURL"
            - type: "null"
      additionalProperties: false
      type: object
      required:
        - type
      title: ChatMessageContent
    ChatMessageContentImageURL:
      properties:
        url:
          type: string
          title: Url
          description:
            Image link or base64 data URI. ``mm_file://{file_id}`` is also
            accepted for assets uploaded via the Files API.
        detail:
          anyOf:
            - type: string
            - type: "null"
          title: Detail
          description:
            Detail level for image understanding. One of ``low`` / ``default``
            / ``high`` (model-specific defaults table). Used to pick a default ``max_long_side_pixel``
            when that field is absent.
        max_long_side_pixel:
          anyOf:
            - type: integer
            - type: "null"
          title: Max Long Side Pixel
          description:
            Per-image cap on the long side after resizing. When omitted,
            the model derives a default from ``detail``. Currently honored by MiniMax
            M3 VL preprocessing; other VL models ignore this field. See the M3 VL
            preprocessing spec (2026-05-29) for the full 3-step resize semantics (long-side
            cap → short-side floor at 112 px → hard total-pixel cap).
      additionalProperties: false
      type: object
      required:
        - url
      title: ChatMessageContentImageURL
    ChatMessageContentVideoURL:
      properties:
        url:
          type: string
          title: Url
          description:
            Video link or base64 data URI. ``mm_file://{file_id}`` is accepted
            for assets uploaded via the Files API (recommended for files > 50 MB).
        detail:
          anyOf:
            - type: string
            - type: "null"
          title: Detail
          description:
            Detail level for video understanding. One of ``low`` / ``default``
            / ``high``.
        max_frames:
          anyOf:
            - type: integer
            - type: "null"
          title: Max Frames
        spatial_limit:
          anyOf:
            - type: integer
            - type: "null"
          title: Spatial Limit
        sample_fps:
          anyOf:
            - type: number
            - type: "null"
          title: Sample Fps
          description:
            Frame sampling rate (Kimi K2.5 VL legacy name). For MiniMax
            M3 VL prefer the canonical ``fps`` field.
        fps:
          anyOf:
            - type: number
            - type: "null"
          title: Fps
          description:
            Frame sampling rate, in [0.2, 5] Hz for MiniMax M3 VL (was
            [0.5, 2] in the pre-2026-05-29 spec). Higher values are more sensitive
            to motion at the cost of more tokens; lower values are cheaper but less
            responsive to fast scene changes. Equivalent to ``sample_fps`` on non-M3
            video models.
        max_long_side_pixel:
          anyOf:
            - type: integer
            - type: "null"
          title: Max Long Side Pixel
          description:
            Per-frame cap on the long side after resizing. When omitted,
            the model derives a default from ``detail``. Currently honored by MiniMax
            M3 VL preprocessing; other VL models ignore this field.
      additionalProperties: false
      type: object
      required:
        - url
      title: ChatMessageContentVideoURL
    CompletionRequest:
      properties:
        model:
          type: string
          title: Model
          description: 'The name of the model to use.


            Example: `"accounts/fireworks/models/kimi-k2-instruct-0905"`'
        user:
          anyOf:
            - type: string
            - type: "null"
          title: User
          description:
            A unique identifier representing your end-user, which can help
            monitor and detect abuse.
        prompt_cache_key:
          anyOf:
            - type: string
            - type: "null"
          title: Prompt Cache Key
          description:
            A key used for prompt caching session affinity. Requests with
            the same prompt_cache_key are routed to the same backend to maximize KV
            cache hit rates. This is the preferred field for session affinity (takes
            priority over the 'user' field).
        prompt_cache_isolation_key:
          anyOf:
            - type: string
            - type: "null"
          title: Prompt Cache Isolation Key
          description: Isolation key for prompt caching to separate cache entries.
        raw_output:
          anyOf:
            - type: boolean
            - type: "null"
          title: Raw Output
          description: Return raw output from the model.
          default: false
        perf_metrics_in_response:
          anyOf:
            - type: boolean
            - type: "null"
          title: Perf Metrics In Response
          description:
            "Whether to include performance metrics in the response body.


            **Non-streaming requests:** Performance metrics are always included in
            response headers (e.g., `fireworks-prompt-tokens`, `fireworks-server-time-to-first-token`).
            Setting this to `true` additionally includes the same metrics in the response
            body under the `perf_metrics` field.


            **Streaming requests:** Performance metrics are only included in the response
            body under the `perf_metrics` field in the final chunk (when `finish_reason`
            is set). This is because headers may not be accessible during streaming.


            The response body `perf_metrics` field contains the following metrics:


            **Basic Metrics (all deployments):**


            - `prompt-tokens`: Number of tokens in the prompt

            - `cached-prompt-tokens`: Number of cached prompt tokens

            - `server-time-to-first-token`: Time from request start to first token
            (in seconds)

            - `server-processing-time`: Total processing time (in seconds, only for
            completed requests)


            **Predicted Outputs Metrics:**


            - `speculation-prompt-tokens`: Number of speculative prompt tokens

            - `speculation-prompt-matched-tokens`: Number of matched speculative prompt
            tokens (for completed requests)


            **Dedicated Deployment Only Metrics:**


            - `speculation-generated-tokens`: Number of speculative generated tokens
            (for completed requests)

            - `speculation-acceptance`: Speculation acceptance rates by position

            - `backend-host`: Hostname of the backend server

            - `num-concurrent-requests`: Number of concurrent requests

            - `deployment`: Deployment name

            - `tokenizer-queue-duration`: Time spent in tokenizer queue

            - `tokenizer-duration`: Time spent in tokenizer

            - `prefill-queue-duration`: Time spent in prefill queue

            - `prefill-duration`: Time spent in prefill

            - `generation-queue-duration`: Time spent in generation queue

            - `generation-duration`: Time spent in generation"
          default: false
        stream:
          anyOf:
            - type: boolean
            - type: "null"
          title: Stream
          description:
            "Whether to stream back partial progress. If set, tokens will
            be sent as data-only [server-sent events](https://developer.mozilla.org/en-US/docs/Web/API/Server-sent_events/Using_server-sent_events#Event_stream_format)
            as they become available, with the stream terminated by a `data: [DONE]`
            message."
          default: false
        stream_options:
          anyOf:
            - $ref: "#/components/schemas/StreamOptions"
            - type: "null"
          description:
            Options for streaming responses. Only valid with ``stream=true``.
            Fireworks includes a final SSE chunk carrying usage totals by default;
            set ``include_usage=false`` to opt out. The ``buffer_tokens`` / ``buffer_ms``
            / ``buffer_mode`` fields coalesce SSE chunks by token count and/or time
            (vLLM backend; override the deployment default).
        n:
          type: integer
          title: N
          description: "How many completions to generate for each prompt.


            **Note:** Because this parameter generates many completions, it can quickly
            consume your token quota. Use carefully and ensure that you have reasonable
            settings for `max_tokens` and `stop`.


            Required range: `1 <= x <= 128`


            Example: `1`"
          default: 1
        service_tier:
          type: string
          enum:
            - auto
            - default
            - flex
            - priority
          title: Service Tier
          description:
            The service tier to use for the request. Specifies the processing
            type used for serving the request. Only "priority" is supported, while
            all other values will be treated as "default" tier.
          default: default
        stop:
          anyOf:
            - type: string
            - items:
                type: string
              type: array
            - type: "null"
          title: Stop
          description:
            Up to 4 sequences where the API will stop generating further
            tokens. The returned text will NOT contain the stop sequence.
        max_tokens:
          anyOf:
            - type: integer
            - type: "null"
          title: Max Tokens
          description:
            The maximum number of tokens to generate in the completion.
            If the token count of your prompt plus max_tokens exceeds the model's
            context length, the behavior depends on context_length_exceeded_behavior.
            By default, max_tokens will be lowered to fit in the context window instead
            of returning an error.
        max_completion_tokens:
          anyOf:
            - type: integer
            - type: "null"
          title: Max Completion Tokens
          description: Alias for max_tokens. Cannot be specified together with max_tokens.
        temperature:
          anyOf:
            - type: number
            - type: "null"
          title: Temperature
          description:
            "What sampling temperature to use, between 0 and 2. Higher
            values like 0.8 will make the output more random, while lower values like
            0.2 will make it more focused and deterministic.


            We generally recommend altering this or top_p but not both.


            Required range: `0 <= x <= 2`


            Example: `1`"
        top_k:
          anyOf:
            - type: integer
            - type: "null"
          title: Top K
          description:
            "Top-k sampling is another sampling method where the k most
            probable next tokens are filtered and the probability mass is redistributed
            among only those k next tokens. The value of k controls the number of
            candidates for the next token at each step during text generation. Must
            be between 0 and 100.


            Required range: `0 <= x <= 100`


            Example: `50`"
        top_p:
          anyOf:
            - type: number
            - type: "null"
          title: Top P
          description:
            "An alternative to sampling with temperature, called nucleus
            sampling, where the model considers the results of the tokens with top_p
            probability mass. So 0.1 means only the tokens comprising the top 10%
            probability mass are considered.


            We generally recommend altering this or temperature but not both.


            Required range: `0 <= x <= 1`


            Example: `1`"
        min_p:
          anyOf:
            - type: number
            - type: "null"
          title: Min P
          description:
            "Minimum probability threshold for token selection. Only tokens
            with probability >= min_p are considered for selection. This is an alternative
            to `top_p` and `top_k` sampling.


            Required range: `0 <= x <= 1`"
        typical_p:
          anyOf:
            - type: number
            - type: "null"
          title: Typical P
          description:
            "Typical-p sampling is an alternative to nucleus sampling.
            It considers the most typical tokens whose cumulative probability is at
            most typical_p.


            Required range: `0 <= x <= 1`"
        frequency_penalty:
          anyOf:
            - type: number
            - type: "null"
          title: Frequency Penalty
          description:
            "Number between -2.0 and 2.0. Positive values penalize new
            tokens based on their existing frequency in the text so far, decreasing
            the model's likelihood to repeat the same line verbatim.


            Reasonable value is around 0.1 to 1 if the aim is to just reduce repetitive
            samples somewhat. If the aim is to strongly suppress repetition, then
            one can increase the coefficients up to 2, but this can noticeably degrade
            the quality of samples. Negative values can be used to increase the likelihood
            of repetition.


            See also `presence_penalty` for penalizing tokens that have at least one
            appearance at a fixed rate.


            OpenAI compatible (follows OpenAI's conventions for handling token frequency
            and repetition penalties).


            Required range: `-2 <= x <= 2`"
        presence_penalty:
          anyOf:
            - type: number
            - type: "null"
          title: Presence Penalty
          description:
            "Number between -2.0 and 2.0. Positive values penalize new
            tokens based on whether they appear in the text so far, increasing the
            model's likelihood to talk about new topics.


            Reasonable value is around 0.1 to 1 if the aim is to just reduce repetitive
            samples somewhat. If the aim is to strongly suppress repetition, then
            one can increase the coefficients up to 2, but this can noticeably degrade
            the quality of samples. Negative values can be used to increase the likelihood
            of repetition.


            See also `frequency_penalty` for penalizing tokens at an increasing rate
            depending on how often they appear.


            OpenAI compatible (follows OpenAI's conventions for handling token frequency
            and repetition penalties).


            Required range: `-2 <= x <= 2`"
        repetition_penalty:
          anyOf:
            - type: number
            - type: "null"
          title: Repetition Penalty
          description:
            "Applies a penalty to repeated tokens to discourage or encourage
            repetition. A value of `1.0` means no penalty, allowing free repetition.
            Values above `1.0` penalize repetition, reducing the likelihood of repeating
            tokens. Values between `0.0` and `1.0` reward repetition, increasing the
            chance of repeated tokens. For a good balance, a value of `1.2` is often
            recommended. Note that the penalty is applied to both the generated output
            and the prompt in decoder-only models.


            Required range: `0 <= x <= 2`"
        mirostat_target:
          anyOf:
            - type: number
            - type: "null"
          title: Mirostat Target
          description:
            "Defines the target perplexity for the Mirostat algorithm.
            Perplexity measures the unpredictability of the generated text, with higher
            values encouraging more diverse and creative outputs, while lower values
            prioritize predictability and coherence. The algorithm dynamically adjusts
            the token selection to maintain this target during text generation.


            If not specified, Mirostat sampling is disabled."
        mirostat_lr:
          anyOf:
            - type: number
            - type: "null"
          title: Mirostat Lr
          description:
            Specifies the learning rate for the Mirostat sampling algorithm,
            which controls how quickly the model adjusts its token distribution to
            maintain the target perplexity. A smaller value slows down the adjustments,
            leading to more stable but gradual shifts, while higher values speed up
            corrections at the cost of potential instability.
        seed:
          anyOf:
            - type: integer
            - type: "null"
          title: Seed
          description: Random seed for deterministic sampling.
        logprobs:
          anyOf:
            - type: integer
            - type: boolean
            - type: "null"
          title: Logprobs
          description:
            "Include log probabilities in the response. This accepts either
            a boolean or an integer:


            If set to `true`, log probabilities are included and the number of alternatives
            can be controlled via `top_logprobs` (OpenAI-compatible behavior).


            If set to an integer N, include log probabilities for up to N most likely
            tokens per position in the legacy format. N must be between 0 and the
            deployment's `--max-logprobs` limit (5 by default).


            The API will always return the logprob of the sampled token, so there
            may be up to `logprobs+1` elements in the response when an integer is
            used."
        top_logprobs:
          anyOf:
            - type: integer
            - type: "null"
          title: Top Logprobs
          description:
            "An integer specifying the number of most likely tokens to
            return at each token position, each with an associated log probability.
            Must be between 0 and the deployment's `--max-logprobs` limit (5 by default).


            When `logprobs` is set, `top_logprobs` can be used to modify how many
            top log probabilities are returned. If `top_logprobs` is not set, the
            API will return up to `logprobs` tokens per position."
        sampling_mask:
          anyOf:
            - type: string
              enum:
                - count
                - non_zero_list
                - non_zero_buffer
            - type: "null"
          title: Sampling Mask
          description:
            Opt-in sampling mask metadata for generated tokens. When set
            to `"count"`, each generated token in the new logprobs format includes
            the number of token logits still eligible for sampling after filters such
            as top_p and top_k are applied. `"non_zero_list"` additionally returns
            active token IDs in `sampling_mask`; `"non_zero_buffer"` additionally
            returns a base64-encoded little-endian uint32 buffer of active token IDs.
            Non-zero payloads are omitted for positions with more active tokens than
            1000.
        echo:
          anyOf:
            - type: boolean
            - type: "null"
          title: Echo
          description: Echo back the prompt in addition to the completion.
          default: false
        echo_last:
          anyOf:
            - type: integer
            - type: "null"
          title: Echo Last
          description:
            Echo back the last N tokens of the prompt in addition to the
            completion. This is useful for obtaining logprobs of the prompt suffix
            but without transferring too much data. Passing `echo_last=len(prompt)`
            is the same as `echo=True`
        ignore_eos:
          type: boolean
          title: Ignore Eos
          description:
            This setting controls whether the model should ignore the End
            of Sequence (EOS) token. When set to `True`, the model will continue generating
            tokens even after the EOS token is produced. By default, it stops when
            the EOS token is reached.
          default: false
        context_length_exceeded_behavior:
          type: string
          enum:
            - error
            - truncate
          title: Context Length Exceeded Behavior
          description:
            "What to do if the token count of prompt plus `max_tokens`
            exceeds the model's context window.


            Passing `truncate` limits the `max_tokens` to at most `context_window_length
            - prompt_length`. This is the default.


            Passing `error` would trigger a request error.


            The default of `'truncate'` is selected as it allows to ask for high
            `max_tokens` value while respecting the context window length without
            having to do client-side prompt tokenization.


            Note, that it differs from OpenAI's behavior that matches that of `error`."
          default: truncate
        response_format:
          anyOf:
            - $ref: "#/components/schemas/ResponseFormat"
            - type: "null"
          description:
            'Allows to force the model to produce specific output format.


            Setting to `{ "type": "json_object" }` enables JSON mode, which guarantees
            the message the model generates is valid JSON.


            If `"type"` is `"json_schema"`, a JSON schema must be provided. E.g.,
            `response_format = {"type": "json_schema", "json_schema": <json_schema>}`.


            Important: when using JSON mode, it''s crucial to also instruct the model
            to produce JSON via a system or user message. Without this, the model
            may generate an unending stream of whitespace until the generation reaches
            the token limit, resulting in a long-running and seemingly "stuck" request.


            Also note that the message content may be partially cut off if `finish_reason="length"`,
            which indicates the generation exceeded `max_tokens` or the conversation
            exceeded the max context length. In this case the return value might not
            be a valid JSON.'
        logit_bias:
          anyOf:
            - additionalProperties:
                type: number
              type: object
            - type: "null"
          title: Logit Bias
          description:
            Modify the likelihood of specified tokens appearing in the
            completion. Accepts a json object that maps tokens (specified by their
            token ID in the tokenizer) to an associated bias value from -100 to 100.
            Mathematically, the bias is added to the logits generated by the model
            prior to sampling.
        speculation:
          anyOf:
            - type: string
            - items:
                type: integer
              type: array
            - type: "null"
          title: Speculation
          description: Speculative decoding prompt or token IDs to speed up generation.
        prediction:
          anyOf:
            - $ref: "#/components/schemas/PredictedOutput"
            - type: string
            - type: "null"
          title: Prediction
          description:
            OpenAI-compatible predicted output for speculative decoding.
            Can be a PredictedOutput object or a simple string. Automatically transformed
            to speculation.
        metadata:
          anyOf:
            - additionalProperties:
                type: string
              type: object
            - type: "null"
          title: Metadata
          description: Additional metadata to store with the request for tracing/distillation.
        reasoning_effort:
          anyOf:
            - type: string
              enum:
                - low
                - medium
                - high
                - xhigh
                - max
                - none
                - adaptive
            - type: integer
            - type: boolean
            - type: "null"
          title: Reasoning Effort
          description:
            'Controls reasoning behavior for supported models. When enabled,
            the model''s reasoning appears in the `reasoning_content` field of the
            response, separate from the final answer in `content`.


            **Accepted values:**


            - **String** (OpenAI-compatible): `''low''`, `''medium''`, `''high''`,
            or `''max''` to enable reasoning with varying effort levels; `''none''`
            to disable reasoning.

            - **Boolean** (Fireworks extension): `true` to enable reasoning, `false`
            to disable it.

            - **Integer** (Fireworks extension): A positive integer to set a hard
            token limit on reasoning output. Integer values enable the model''s normal
            medium-style thinking behavior and force the model to end its thinking
            phase after at most that many generated thinking tokens.


            **Important:** Boolean values are normalized internally: `true` becomes
            `''medium''`, and `false` becomes `''none''`. This normalization happens
            before model-specific validation, so if a model doesn''t support `''none''`,
            passing `false` will produce an error referencing `''none''`.


            **Model-specific behavior:**


            - **Qwen3**: Grammar-based reasoning on the reasoning-enabled `qwen3`/`qwen3p5`
            conversation styles. Older chat-mode Qwen3 deployments may opt into `qwen3-no-thinking`,
            which disables reasoning support. For reasoning-enabled styles, use `''none''`
            or `false` to disable. Supports integer token limits to cap reasoning
            output. `''low''`, `''medium''`, and `''high''` keep their model-specific
            behavior and are not hard budgets.

            - **MiniMax M2**: Reasoning is required (always on). Defaults to `''medium''`
            when omitted. Accepts only string `reasoning_effort`: `''low''`, `''medium''`,
            or `''high''`. `''none''` and boolean values are rejected.

            - **DeepSeek V3.1**: Binary on/off reasoning. Default reasoning off (matches
            chat template). Use `true`, `''low''`, `''medium''`, or `''high''` to
            enable; `''none''` or `false` to disable.

            - **DeepSeek V3.2**: Binary on/off reasoning. Default reasoning on. Use
            `''none''` or `false` to disable; effort levels and integers have no additional
            effect.

            - **DeepSeek V4**: Accepts `''none''`, `''low''`, `''medium''`, `''high''`,
            `''xhigh''`, and `''max''`. Default reasoning on (`''high''`). `''xhigh''`
            is silently promoted to `''max''`. `''max''` prepends a thorough-reasoning
            preamble; `''high''` enables thinking. `''low''` and `''medium''` are
            silently promoted to `''high''`. `''none''` or `false` disables thinking.

            - **GLM 4.5, GLM 4.5 Air, GLM 4.6, GLM 4.7, GLM 5.1**: Binary on/off reasoning.
            Default reasoning on. Use `''none''` or `false` to disable; effort levels
            and integers have no additional effect.

            - **GLM 5.2**: Two thinking tiers, `High` and `Max` (rendered as a `Reasoning
            Effort:` system line). `''high''` selects High; `''low''` and `''medium''`
            are collapsed to `''high''`; `''max''` and `''xhigh''` select Max; when
            omitted, the model default (`Max`) applies. `''none''` or `false` disables
            thinking.

            - **Harmony (OpenAI GPT-OSS 120B, GPT-OSS 20B)**: Accepts only `''low''`,
            `''medium''`, or `''high''`. Does not support `''none''`, `false`, or
            integer values — using these will return an error (e.g., "Invalid reasoning
            effort: none"). When omitted, defaults to `''medium''`. Lower effort produces
            faster responses with shorter reasoning.'
        reasoning_history:
          anyOf:
            - type: string
              enum:
                - disabled
                - interleaved
                - preserved
            - type: "null"
          title: Reasoning History
          description:
            "Controls how historical assistant reasoning content is included
            in the prompt for multi-turn conversations.


            **Accepted values:**


            - `null`: Use model/template default behavior (for **GLM-4.7**, the model/template
            default is `'interleaved'`, i.e. historical reasoning is cleared by
            default)

            - `'disabled'`: Strip `reasoning_content` from all messages before prompt
            construction

            - `'interleaved'`: Strip `reasoning_content` from messages up to (and
            including) the last user message

            - `'preserved'`: Preserve historical `reasoning_content` across the
            conversation


            **Model support:**


            | Model | Default | Supported values |

            | --- | --- | --- |

            | Kimi K2.7 | `'preserved'` | `'disabled'`, `'interleaved'`, `'preserved'`
            |

            | Kimi K2.6 | `'interleaved'` | `'disabled'`, `'interleaved'`, `'preserved'`
            |

            | Kimi K2 Instruct | `'preserved'` | `'disabled'`, `'interleaved'`,
            `'preserved'` |

            | MiniMax M2 | `'interleaved'` | `'disabled'`, `'interleaved'` |

            | GLM-5.2 | `'interleaved'` | `'disabled'`, `'interleaved'`, `'preserved'`
            |

            | GLM-4.7 | `'interleaved'` | `'disabled'`, `'interleaved'`, `'preserved'`
            |

            | GLM-4.6 | `'interleaved'` | `'disabled'`, `'interleaved'` |

            | Qwen 3.6 | `'preserved'` | `'disabled'`, `'preserved'` |

            | DeepSeek V4 | `'interleaved'` | `'interleaved'` |


            For other models, refer to the model provider's documentation.


            **Note:** This parameter controls prompt formatting only. To disable reasoning
            computation entirely, use `reasoning_effort='none'`."
        thinking:
          anyOf:
            - $ref: "#/components/schemas/ThinkingConfigEnabled"
            - $ref: "#/components/schemas/ThinkingConfigDisabled"
            - $ref: "#/components/schemas/ThinkingConfigAdaptive"
            - type: "null"
          title: Thinking
          description:
            'Configuration for enabling extended thinking (Anthropic-compatible
            format). This is an alternative to `reasoning_effort` for controlling
            reasoning behavior.


            **Format:**


            - `{"type": "enabled"}` - Enable thinking (equivalent to `reasoning_effort:
            true`)

            - `{"type": "enabled", "budget_tokens": <int>}` - Enable thinking with
            a token budget (equivalent to `reasoning_effort: <int>`). Must be >= 1024.

            - `{"type": "enabled", "keep": "all"}` - Enable thinking and preserve
            all historical reasoning content in the prompt (equivalent to `reasoning_history:
            "preserved"`).

            - `{"type": "disabled"}` - Disable thinking (equivalent to `reasoning_effort:
            "none"`)


            **Precedence with `reasoning_effort`:** `thinking.effort` (when set) overrides
            `reasoning_effort`; otherwise, for `type=enabled`, `reasoning_effort`
            is used as the effort level. `type=disabled` always disables thinking.'
        return_token_ids:
          anyOf:
            - type: boolean
            - type: "null"
          title: Return Token Ids
          description: Return token IDs alongside text to avoid retokenization drift.
          default: false
        prompt:
          anyOf:
            - type: string
            - items:
                type: string
              type: array
            - items:
                type: integer
              type: array
            - items:
                items:
                  type: integer
                type: array
              type: array
          title: Prompt
          description: "The prompt to generate completions for.


            It can be a single string or an array of strings.


            It can also be an array of integers or an array of integer arrays, which
            allows to pass already tokenized prompt.


            If multiple prompts are specified, several choices with corresponding
            `index` will be returned in the output."
        images:
          anyOf:
            - items:
                type: string
              type: array
            - items:
                items:
                  type: string
                type: array
              type: array
            - type: "null"
          title: Images
          description:
            "The list of base64 encoded images for visual language completition
            generation.


            They should be formatted as MIME_TYPE,<base64 encoded str>


            eg. data:image/jpeg;base64,<base64 encoded str>


            Additionally, the number of images provided should match the number of
            image placeholder tokens in the prompt (string prompts: '<image>' or
            model-specific pads such as '<|image_pad|>'; tokenized prompts: one
            image pad token ID per image, unexpanded)."
      additionalProperties: false
      type: object
      required:
        - prompt
        - model
      title: CompletionRequest
    EmbeddingRequest:
      properties:
        model:
          type: string
          title: Model
          description: The name of the model to use.
        user:
          anyOf:
            - type: string
            - type: "null"
          title: User
          description:
            A unique identifier representing your end-user, which can help
            monitor and detect abuse.
        prompt_cache_key:
          anyOf:
            - type: string
            - type: "null"
          title: Prompt Cache Key
          description:
            A key used for prompt caching session affinity. Requests with
            the same prompt_cache_key are routed to the same backend to maximize KV
            cache hit rates. This is the preferred field for session affinity (takes
            priority over the 'user' field).
        prompt_cache_isolation_key:
          anyOf:
            - type: string
            - type: "null"
          title: Prompt Cache Isolation Key
          description: Isolation key for prompt caching to separate cache entries.
        raw_output:
          anyOf:
            - type: boolean
            - type: "null"
          title: Raw Output
          description: Return raw output from the model.
          default: false
        perf_metrics_in_response:
          anyOf:
            - type: boolean
            - type: "null"
          title: Perf Metrics In Response
          description:
            "Whether to include performance metrics in the response body.


            **Non-streaming requests:** Performance metrics are always included in
            response headers (e.g., `fireworks-prompt-tokens`, `fireworks-server-time-to-first-token`).
            Setting this to `true` additionally includes the same metrics in the response
            body under the `perf_metrics` field.


            **Streaming requests:** Performance metrics are only included in the response
            body under the `perf_metrics` field in the final chunk (when `finish_reason`
            is set). This is because headers may not be accessible during streaming.


            The response body `perf_metrics` field contains the following metrics:


            **Basic Metrics (all deployments):**


            - `prompt-tokens`: Number of tokens in the prompt

            - `cached-prompt-tokens`: Number of cached prompt tokens

            - `server-time-to-first-token`: Time from request start to first token
            (in seconds)

            - `server-processing-time`: Total processing time (in seconds, only for
            completed requests)


            **Predicted Outputs Metrics:**


            - `speculation-prompt-tokens`: Number of speculative prompt tokens

            - `speculation-prompt-matched-tokens`: Number of matched speculative prompt
            tokens (for completed requests)


            **Dedicated Deployment Only Metrics:**


            - `speculation-generated-tokens`: Number of speculative generated tokens
            (for completed requests)

            - `speculation-acceptance`: Speculation acceptance rates by position

            - `backend-host`: Hostname of the backend server

            - `num-concurrent-requests`: Number of concurrent requests

            - `deployment`: Deployment name

            - `tokenizer-queue-duration`: Time spent in tokenizer queue

            - `tokenizer-duration`: Time spent in tokenizer

            - `prefill-queue-duration`: Time spent in prefill queue

            - `prefill-duration`: Time spent in prefill

            - `generation-queue-duration`: Time spent in generation queue

            - `generation-duration`: Time spent in generation"
          default: false
        input:
          anyOf:
            - type: string
            - items:
                type: string
              type: array
            - items:
                items:
                  type: integer
                type: array
              type: array
            - items:
                type: integer
              type: array
            - additionalProperties: true
              type: object
            - items:
                additionalProperties: true
                type: object
              type: array
          title: Input
        prompt_template:
          anyOf:
            - type: string
            - type: "null"
          title: Prompt Template
        encoding_format:
          type: string
          enum:
            - float
            - base64
          title: Encoding Format
          default: float
        dimensions:
          anyOf:
            - type: integer
            - type: "null"
          title: Dimensions
        return_logits:
          anyOf:
            - items:
                type: integer
              type: array
            - type: "null"
          title: Return Logits
        normalize:
          anyOf:
            - type: boolean
            - type: "null"
          title: Normalize
          default: false
        fanout_size:
          anyOf:
            - type: integer
            - type: "null"
          title: Fanout Size
      additionalProperties: false
      type: object
      required:
        - input
        - model
      title: EmbeddingRequest
    FunctionNameSpec:
      properties:
        name:
          type: string
          title: Name
      additionalProperties: false
      type: object
      required:
        - name
      title: FunctionNameSpec
    FunctionSelection:
      properties:
        type:
          type: string
          const: function
          title: Type
        function:
          anyOf:
            - $ref: "#/components/schemas/FunctionNameSpec"
            - type: "null"
      additionalProperties: false
      type: object
      required:
        - type
      title: FunctionSelection
    HTTPValidationError:
      properties:
        detail:
          items:
            $ref: "#/components/schemas/ValidationError"
          type: array
          title: Detail
      type: object
      title: HTTPValidationError
    PredictedOutput:
      properties:
        content:
          anyOf:
            - type: string
            - items:
                $ref: "#/components/schemas/ChatMessageContent"
              type: array
          title: Content
        type:
          type: string
          const: content
          title: Type
          default: content
      additionalProperties: false
      type: object
      required:
        - content
      title: PredictedOutput
      description: OpenAI-compatible struct for the "speculation" field.
    RerankRequestBody:
      properties:
        model:
          anyOf:
            - type: string
            - type: "null"
          title: Model
        query:
          type: string
          title: Query
        documents:
          items:
            type: string
          type: array
          title: Documents
        top_n:
          anyOf:
            - type: integer
            - type: "null"
          title: Top N
        return_documents:
          type: boolean
          title: Return Documents
          default: true
        task:
          anyOf:
            - type: string
            - type: "null"
          title: Task
      type: object
      required:
        - query
        - documents
      title: RerankRequestBody
    ResponseFormat:
      properties:
        type:
          type: string
          enum:
            - json_object
            - json_schema
            - grammar
            - text
          title: Type
        schema:
          anyOf:
            - additionalProperties: true
              type: object
            - type: string
            - type: "null"
          title: Schema
        grammar:
          anyOf:
            - type: string
            - type: "null"
          title: Grammar
        json_schema:
          anyOf:
            - additionalProperties: true
              type: object
            - type: string
            - type: "null"
          title: Json Schema
      additionalProperties: false
      type: object
      required:
        - type
      title: ResponseFormat
    StreamOptions:
      properties:
        include_usage:
          anyOf:
            - type: boolean
            - type: "null"
          title: Include Usage
          description:
            "Whether to include a trailing SSE chunk with usage totals
            (with an empty `choices` array). Unlike the OpenAI spec, Fireworks includes
            usage by default for streaming responses; set this to `false` to opt out.
            When emitted, usage rides a separate final chunk (before `data: [DONE]`),
            not the chunk carrying `finish_reason`."
        include_internal_content:
          anyOf:
            - type: boolean
            - type: "null"
          title: Include Internal Content
          description:
            When true, include an `internal_content` object (currently
            `token_id`) inside each streaming delta — equivalent to `return_token_ids`,
            emitted under `choices[].delta.internal_content`. Omitted entirely from
            response chunks when false.
          default: false
        buffer_tokens:
          anyOf:
            - type: integer
              minimum: 0.0
            - type: "null"
          title: Buffer Tokens
          description:
            Coalesce streaming SSE chunks until this many text deltas (~tokens)
            accumulate before flushing a merged chunk. 0 disables the token threshold.
            Honored only on the vLLM backend; overrides the deployment default when
            set.
        buffer_ms:
          anyOf:
            - type: number
              minimum: 0.0
            - type: "null"
          title: Buffer Ms
          description:
            Coalesce streaming SSE chunks for up to this many milliseconds
            before flushing a merged chunk. 0 disables the time threshold. Honored
            only on the vLLM backend; overrides the deployment default when set.
        buffer_mode:
          anyOf:
            - type: string
              enum:
                - any
                - all
            - type: "null"
          title: Buffer Mode
          description:
            "When both buffer_tokens and buffer_ms are set: 'any' flushes
            when either threshold is reached; 'all' flushes only when both are.
            Overrides the deployment default when set; defaults to 'any'."
      additionalProperties: false
      type: object
      title: StreamOptions
    TgiGenerateParameters:
      properties:
        details:
          type: boolean
          title: Details
          default: false
        max_new_tokens:
          anyOf:
            - type: integer
            - type: "null"
          title: Max New Tokens
        repetition_penalty:
          anyOf:
            - type: number
            - type: "null"
          title: Repetition Penalty
        return_full_text:
          type: boolean
          title: Return Full Text
          default: false
        stop:
          anyOf:
            - items:
                type: string
              type: array
            - type: "null"
          title: Stop
        temperature:
          anyOf:
            - type: number
            - type: "null"
          title: Temperature
        top_k:
          anyOf:
            - type: integer
            - type: "null"
          title: Top K
        top_p:
          anyOf:
            - type: number
            - type: "null"
          title: Top P
        truncate:
          anyOf:
            - type: integer
            - type: "null"
          title: Truncate
        typical_p:
          anyOf:
            - type: number
            - type: "null"
          title: Typical P
      additionalProperties: false
      type: object
      title: TgiGenerateParameters
    TgiGenerateRequest:
      properties:
        model:
          type: string
          title: Model
          description: The name of the model to use.
        user:
          anyOf:
            - type: string
            - type: "null"
          title: User
          description:
            A unique identifier representing your end-user, which can help
            monitor and detect abuse.
        prompt_cache_key:
          anyOf:
            - type: string
            - type: "null"
          title: Prompt Cache Key
          description:
            A key used for prompt caching session affinity. Requests with
            the same prompt_cache_key are routed to the same backend to maximize KV
            cache hit rates. This is the preferred field for session affinity (takes
            priority over the 'user' field).
        prompt_cache_isolation_key:
          anyOf:
            - type: string
            - type: "null"
          title: Prompt Cache Isolation Key
          description: Isolation key for prompt caching to separate cache entries.
        raw_output:
          anyOf:
            - type: boolean
            - type: "null"
          title: Raw Output
          description: Return raw output from the model.
          default: false
        perf_metrics_in_response:
          anyOf:
            - type: boolean
            - type: "null"
          title: Perf Metrics In Response
          description:
            "Whether to include performance metrics in the response body.


            **Non-streaming requests:** Performance metrics are always included in
            response headers (e.g., `fireworks-prompt-tokens`, `fireworks-server-time-to-first-token`).
            Setting this to `true` additionally includes the same metrics in the response
            body under the `perf_metrics` field.


            **Streaming requests:** Performance metrics are only included in the response
            body under the `perf_metrics` field in the final chunk (when `finish_reason`
            is set). This is because headers may not be accessible during streaming.


            The response body `perf_metrics` field contains the following metrics:


            **Basic Metrics (all deployments):**


            - `prompt-tokens`: Number of tokens in the prompt

            - `cached-prompt-tokens`: Number of cached prompt tokens

            - `server-time-to-first-token`: Time from request start to first token
            (in seconds)

            - `server-processing-time`: Total processing time (in seconds, only for
            completed requests)


            **Predicted Outputs Metrics:**


            - `speculation-prompt-tokens`: Number of speculative prompt tokens

            - `speculation-prompt-matched-tokens`: Number of matched speculative prompt
            tokens (for completed requests)


            **Dedicated Deployment Only Metrics:**


            - `speculation-generated-tokens`: Number of speculative generated tokens
            (for completed requests)

            - `speculation-acceptance`: Speculation acceptance rates by position

            - `backend-host`: Hostname of the backend server

            - `num-concurrent-requests`: Number of concurrent requests

            - `deployment`: Deployment name

            - `tokenizer-queue-duration`: Time spent in tokenizer queue

            - `tokenizer-duration`: Time spent in tokenizer

            - `prefill-queue-duration`: Time spent in prefill queue

            - `prefill-duration`: Time spent in prefill

            - `generation-queue-duration`: Time spent in generation queue

            - `generation-duration`: Time spent in generation"
          default: false
        inputs:
          type: string
          title: Inputs
        parameters:
          anyOf:
            - $ref: "#/components/schemas/TgiGenerateParameters"
            - type: "null"
        stream:
          anyOf:
            - type: boolean
            - type: "null"
          title: Stream
          default: false
      additionalProperties: false
      type: object
      required:
        - inputs
        - model
      title: TgiGenerateRequest
    ThinkingConfigAdaptive:
      properties:
        type:
          type: string
          const: adaptive
          title: Type
      additionalProperties: false
      type: object
      required:
        - type
      title: ThinkingConfigAdaptive
      description:
        'Configuration that lets the model decide whether to think (MiniMax
        M3).


        Per the M3 API spec (§1.4), `{"type": "adaptive"}` is M3''s default — the
        model

        decides whether to emit a thinking phase. No forced first token is applied.

        Currently accepted only by MiniMax M3; other model families reject it.'
    ThinkingConfigDisabled:
      properties:
        type:
          type: string
          const: disabled
          title: Type
      additionalProperties: false
      type: object
      required:
        - type
      title: ThinkingConfigDisabled
      description:
        Configuration for disabling extended thinking (Anthropic-compatible
        format).
    ThinkingConfigEnabled:
      properties:
        type:
          type: string
          const: enabled
          title: Type
          default: enabled
        budget_tokens:
          anyOf:
            - type: integer
            - type: "null"
          title: Budget Tokens
          description:
            Determines how many tokens the model can use for its internal
            reasoning process. Larger budgets can enable more thorough analysis for
            complex problems, improving response quality. Must be >= 1024 if specified.
        keep:
          anyOf:
            - type: string
              const: all
            - type: "null"
          title: Keep
          description:
            'Controls whether historical reasoning content is preserved
            in the prompt. When set to `"all"`, all previous assistant turns'' reasoning_content
            is included in the rendered prompt (equivalent to `reasoning_history:
            "preserved"`). When omitted (null), the model uses its default history
            behavior. Only valid when `type` is `"enabled"`.'
        budget_end_str:
          anyOf:
            - type: string
            - type: "null"
          title: Budget End Str
          description:
            Natural-language transition phrase that the model is forced
            to emit just before the end-thinking token (`</think>`) when `budget_tokens`
            is exhausted. This produces a more natural conclusion than a hard token
            slam (matches vLLM's `reasoning_end_str` behavior). Defaults to a built-in
            phrase. Set to "" to disable the bridge and force `</think>` immediately.
            Only meaningful when `budget_tokens` is set.
        effort:
          anyOf:
            - type: string
              enum:
                - low
                - medium
                - high
                - xhigh
                - max
                - none
                - adaptive
            - type: integer
            - type: boolean
            - type: "null"
          title: Effort
          description:
            Reasoning effort level (Kimi/Moonshot spec). Accepts the same
            values as top-level `reasoning_effort`; when set it takes precedence over
            (overwrites) `reasoning_effort`.
      additionalProperties: false
      type: object
      title: ThinkingConfigEnabled
      description:
        Configuration for enabling extended thinking (Anthropic-compatible
        format).
    ValidationError:
      properties:
        loc:
          items:
            anyOf:
              - type: string
              - type: integer
          type: array
          title: Location
        msg:
          type: string
          title: Message
        type:
          type: string
          title: Error Type
        input:
          title: Input
        ctx:
          type: object
          title: Context
      type: object
      required:
        - loc
        - msg
        - type
      title: ValidationError
    CompletionResponse:
      additionalProperties: false
      description: The response message from a /v1/completions call.
      properties:
        id:
          description: A unique identifier of the response
          title: Id
          type: string
        object:
          default: text_completion
          description: The object type, which is always "text_completion"
          title: Object
          type: string
        created:
          description: The Unix time in seconds when the response was generated
          title: Created
          type: integer
        model:
          description: The model used for the completion
          title: Model
          type: string
        choices:
          description: The list of generated completion choices
          items:
            $ref: "#/components/schemas/Choice"
          title: Choices
          type: array
        usage:
          $ref: "#/components/schemas/UsageInfo"
          description: Usage statistics for the completion
        perf_metrics:
          anyOf:
            - additionalProperties: true
              type: object
            - type: "null"
          default: null
          description: See parameter [perf_metrics_in_response](#body-perf-metrics-in-response)
          title: Perf Metrics
      required:
        - id
        - created
        - model
        - choices
        - usage
      title: CompletionResponse
      type: object
    Choice:
      additionalProperties: false
      description: A completion choice.
      properties:
        index:
          description: The index of the completion choice
          title: Index
          type: integer
        text:
          description: The completion response
          title: Text
          type: string
        logprobs:
          anyOf:
            - $ref: "#/components/schemas/LogProbs"
            - $ref: "#/components/schemas/NewLogProbs"
            - type: "null"
          default: null
          description: The log probabilities of the most likely tokens
          title: Logprobs
        finish_reason:
          anyOf:
            - enum:
                - stop
                - length
                - error
              type: string
            - type: "null"
          default: null
          description:
            The reason the model stopped generating tokens. This will be
            "stop" if the model hit a natural stop point or a provided stop sequence,
            or "length" if the maximum number of tokens specified in the request was
            reached
          title: Finish Reason
        raw_output:
          anyOf:
            - $ref: "#/components/schemas/RawOutput"
            - type: "null"
          default: null
        prompt_token_ids:
          anyOf:
            - items:
                type: integer
              type: array
            - type: "null"
          default: null
          description: Token IDs for the prompt (when return_token_ids=true)
          title: Prompt Token Ids
        token_ids:
          anyOf:
            - items:
                type: integer
              type: array
            - type: "null"
          default: null
          description: Token IDs for the generated completion (when return_token_ids=true)
          title: Token Ids
      required:
        - index
        - text
      title: Choice
      type: object
    LogProbs:
      additionalProperties: false
      description: Legacy log probabilities format
      properties:
        tokens:
          items:
            type: string
          title: Tokens
          type: array
        token_logprobs:
          items:
            type: number
          title: Token Logprobs
          type: array
        top_logprobs:
          anyOf:
            - items:
                additionalProperties:
                  type: number
                type: object
              type: array
            - type: "null"
          title: Top Logprobs
        text_offset:
          items:
            type: integer
          title: Text Offset
          type: array
        token_ids:
          anyOf:
            - items:
                type: integer
              type: array
            - type: "null"
          default: null
          title: Token Ids
      title: LogProbs
      type: object
    NewLogProbs:
      additionalProperties: false
      description: OpenAI-compatible log probabilities format
      properties:
        content:
          items:
            $ref: "#/components/schemas/NewLogProbsContent"
          title: Content
          type: array
      title: NewLogProbs
      type: object
    NewLogProbsContent:
      additionalProperties: false
      properties:
        token:
          title: Token
          type: string
        logprob:
          title: Logprob
          type: number
        sampling_logprob:
          anyOf:
            - type: number
            - type: "null"
          title: Sampling Logprob
        sampling_mask_count:
          anyOf:
            - type: integer
            - type: "null"
          default: null
          title: Sampling Mask Count
        sampling_mask:
          anyOf:
            - items:
                type: integer
              type: array
            - type: string
            - type: "null"
          default: null
          title: Sampling Mask
        bytes:
          items:
            type: integer
          title: Bytes
          type: array
        top_logprobs:
          items:
            $ref: "#/components/schemas/NewLogProbsContentTopLogProbs"
          title: Top Logprobs
          type: array
        token_id:
          title: Token Id
          type: integer
        text_offset:
          title: Text Offset
          type: integer
        last_activation:
          anyOf:
            - type: string
            - type: "null"
          default: null
          title: Last Activation
        routing_matrix:
          anyOf:
            - type: string
            - type: "null"
          default: null
          title: Routing Matrix
      required:
        - token
        - logprob
        - sampling_logprob
        - bytes
        - token_id
        - text_offset
      title: NewLogProbsContent
      type: object
    NewLogProbsContentTopLogProbs:
      additionalProperties: false
      properties:
        token:
          title: Token
          type: string
        logprob:
          title: Logprob
          type: number
        token_id:
          title: Token Id
          type: integer
        bytes:
          items:
            type: integer
          title: Bytes
          type: array
      required:
        - token
        - logprob
        - token_id
      title: NewLogProbsContentTopLogProbs
      type: object
    PromptTokensDetails:
      additionalProperties: false
      properties:
        cached_tokens:
          anyOf:
            - type: integer
            - type: "null"
          default: null
          title: Cached Tokens
      title: PromptTokensDetails
      type: object
    RawOutput:
      additionalProperties: false
      description:
        "Extension of OpenAI that returns low-level interaction of what
        the model

        sees, including the formatted prompt and function calls"
      properties:
        prompt_fragments:
          description:
            Pieces of the prompt (like individual messages) before truncation
            and concatenation. Depending on prompt_truncate_len some of the messages
            might be dropped. Contains a mix of strings to be tokenized and individual
            tokens (if dictated by the conversation template)
          items:
            anyOf:
              - type: string
              - type: integer
          title: Prompt Fragments
          type: array
        prompt_token_ids:
          description: Fully processed prompt as seen by the model
          items:
            type: integer
          title: Prompt Token Ids
          type: array
        completion:
          description:
            Raw completion produced by the model before any tool calls
            are parsed
          title: Completion
          type: string
        completion_token_ids:
          anyOf:
            - items:
                type: integer
              type: array
            - type: "null"
          default: null
          description: Token IDs for the raw completion
          title: Completion Token Ids
        completion_logprobs:
          anyOf:
            - $ref: "#/components/schemas/NewLogProbs"
            - type: "null"
          default: null
          description:
            Log probabilities for the completion. Only populated if logprobs
            is specified in the request
        images:
          anyOf:
            - items:
                type: string
              type: array
            - type: "null"
          default: null
          description: Images in the prompt
          title: Images
        videos:
          anyOf:
            - items:
                type: string
              type: array
            - type: "null"
          default: null
          description: Videos in the prompt
          title: Videos
        grammar:
          anyOf:
            - type: string
            - type: "null"
          default: null
          description:
            Grammar used for constrained decoding, can be either user provided
            (directly or JSON schema) or inferred by the chat template
          title: Grammar
      required:
        - prompt_fragments
        - prompt_token_ids
        - completion
      title: RawOutput
      type: object
    UsageInfo:
      additionalProperties: false
      description: Usage statistics.
      properties:
        prompt_tokens:
          description: The number of tokens in the prompt
          title: Prompt Tokens
          type: integer
        total_tokens:
          description: The total number of tokens used in the request (prompt + completion)
          title: Total Tokens
          type: integer
        completion_tokens:
          anyOf:
            - type: integer
            - type: "null"
          default: null
          description: The number of tokens in the generated completion
          title: Completion Tokens
        prompt_tokens_details:
          anyOf:
            - $ref: "#/components/schemas/PromptTokensDetails"
            - type: "null"
          default: null
          description: Details about prompt tokens, including cached tokens
      required:
        - prompt_tokens
        - total_tokens
      title: UsageInfo
      type: object
    CompletionStreamResponse:
      additionalProperties: false
      description: The streamed response message from a /v1/completions call.
      properties:
        id:
          description: A unique identifier of the response
          title: Id
          type: string
        object:
          default: text_completion
          description: The object type, which is always "text_completion"
          title: Object
          type: string
        created:
          description: The Unix time in seconds when the response was generated
          title: Created
          type: integer
        model:
          description: The model used for the chat completion
          title: Model
          type: string
        choices:
          description: The list of streamed completion choices
          items:
            $ref: "#/components/schemas/CompletionResponseStreamChoice"
          title: Choices
          type: array
        usage:
          anyOf:
            - $ref: "#/components/schemas/UsageInfo"
            - type: "null"
          default: null
        perf_metrics:
          anyOf:
            - additionalProperties: true
              type: object
            - type: "null"
          default: null
          description: See parameter [perf_metrics_in_response](#body-perf-metrics-in-response)
          title: Perf Metrics
      required:
        - id
        - created
        - model
        - choices
      title: CompletionStreamResponse
      type: object
    CompletionResponseStreamChoice:
      additionalProperties: false
      description:
        "A streamed completion choice.\n\nAttributes:\n  index (int): The\
        \ index of the completion choice.\n  text (str): The completion response.\n\
        \  logprobs (float, optional): The log probabilities of the most likely tokens.\n\
        \  finish_reason (str): The reason the model stopped generating tokens. This\
        \ will be \"stop\" if\n    the model hit a natural stop point or a provided\
        \ stop sequence, or\n    \"length\" if the maximum number of tokens specified\
        \ in the request was\n    reached.\n  prompt_token_ids (Optional[List[int]]):\
        \ Token IDs for the prompt (when return_token_ids=true, sent in first chunk)\n\
        \  token_ids (Optional[List[int]]): Token IDs for this chunk (when return_token_ids=true)"
      properties:
        index:
          title: Index
          type: integer
        text:
          title: Text
          type: string
        logprobs:
          anyOf:
            - $ref: "#/components/schemas/LogProbs"
            - $ref: "#/components/schemas/NewLogProbs"
            - type: "null"
          default: null
          title: Logprobs
        finish_reason:
          anyOf:
            - enum:
                - stop
                - length
                - error
              type: string
            - type: "null"
          default: null
          title: Finish Reason
        raw_output:
          anyOf:
            - $ref: "#/components/schemas/RawOutput"
            - type: "null"
          default: null
        prompt_token_ids:
          anyOf:
            - items:
                type: integer
              type: array
            - type: "null"
          default: null
          title: Prompt Token Ids
        token_ids:
          anyOf:
            - items:
                type: integer
              type: array
            - type: "null"
          default: null
          title: Token Ids
      required:
        - index
        - text
      title: CompletionResponseStreamChoice
      type: object
    ChatCompletionResponse:
      additionalProperties: false
      description: The response message from a /v1/chat/completions call.
      properties:
        id:
          description: A unique identifier of the response
          title: Id
          type: string
        object:
          default: chat.completion
          description: The object type, which is always "chat.completion"
          title: Object
          type: string
        created:
          description: The Unix time in seconds when the response was generated
          title: Created
          type: integer
        model:
          description: The model used for the chat completion
          title: Model
          type: string
        choices:
          description: The list of chat completion choices
          items:
            $ref: "#/components/schemas/ChatCompletionResponseChoice"
          title: Choices
          type: array
        usage:
          anyOf:
            - $ref: "#/components/schemas/UsageInfo"
            - type: "null"
          default: null
        perf_metrics:
          anyOf:
            - additionalProperties: true
              type: object
            - type: "null"
          default: null
          description: See parameter [perf_metrics_in_response](#body-perf-metrics-in-response)
          title: Perf Metrics
        prompt_token_ids:
          anyOf:
            - items:
                type: integer
              type: array
            - type: "null"
          default: null
          description: Token IDs for the prompt (when return_token_ids=true)
          title: Prompt Token Ids
      required:
        - id
        - created
        - model
        - choices
      title: ChatCompletionResponse
      type: object
    ChatCompletionResponseChoice:
      additionalProperties: false
      description:
        "A chat completion choice generated by a chat model.\n\nAttributes:\n\
        \  index (int): The index of the chat completion choice.\n  message (ChatMessage):\
        \ The chat completion message.\n  finish_reason (Optional[str]): The reason\
        \ the model stopped generating tokens. This will be \"stop\" if\n    the model\
        \ hit a natural stop point or a provided stop sequence, or\n    \"length\"\
        \ if the maximum number of tokens specified in the request was\n    reached.\n\
        \  token_ids (Optional[List[int]]): Token IDs for the generated message content\
        \ (when return_token_ids=true)"
      properties:
        index:
          title: Index
          type: integer
        message:
          $ref: "#/components/schemas/ChatMessage"
        finish_reason:
          anyOf:
            - type: string
            - type: "null"
          default: null
          title: Finish Reason
        logprobs:
          anyOf:
            - $ref: "#/components/schemas/LogProbs"
            - $ref: "#/components/schemas/NewLogProbs"
            - type: "null"
          default: null
          title: Logprobs
        raw_output:
          anyOf:
            - $ref: "#/components/schemas/RawOutput"
            - type: "null"
          default: null
        token_ids:
          anyOf:
            - items:
                type: integer
              type: array
            - type: "null"
          default: null
          title: Token Ids
      required:
        - index
        - message
      title: ChatCompletionResponseChoice
      type: object
    ChatCompletionStreamResponse:
      additionalProperties: false
      description: The streamed response message from a /v1/chat/completions call.
      properties:
        id:
          description: A unique identifier of the response
          title: Id
          type: string
        object:
          default: chat.completion.chunk
          description: The object type, which is always "chat.completion.chunk"
          title: Object
          type: string
        created:
          description: The Unix time in seconds when the response was generated
          title: Created
          type: integer
        model:
          description: The model used for the chat completion
          title: Model
          type: string
        choices:
          description: The list of streamed chat completion choices
          items:
            $ref: "#/components/schemas/ChatCompletionResponseStreamChoice"
          title: Choices
          type: array
        usage:
          anyOf:
            - $ref: "#/components/schemas/UsageInfo"
            - type: "null"
          default: null
        perf_metrics:
          anyOf:
            - additionalProperties: true
              type: object
            - type: "null"
          default: null
          description: See parameter [perf_metrics_in_response](#body-perf-metrics-in-response)
          title: Perf Metrics
        prompt_token_ids:
          anyOf:
            - items:
                type: integer
              type: array
            - type: "null"
          default: null
          description:
            Token IDs for the prompt (when return_token_ids=true, sent
            in first chunk)
          title: Prompt Token Ids
      required:
        - id
        - created
        - model
        - choices
      title: ChatCompletionStreamResponse
      type: object
    ChatCompletionResponseStreamChoice:
      additionalProperties: false
      description: A streamed chat completion choice.
      properties:
        index:
          description: The index of the chat completion choice
          title: Index
          type: integer
        delta:
          $ref: "#/components/schemas/DeltaMessage"
          description: The message delta
        finish_reason:
          anyOf:
            - enum:
                - stop
                - length
                - function_call
                - tool_calls
              type: string
            - type: "null"
          default: null
          description:
            The reason the model stopped generating tokens. This will be
            "stop" if the model hit a natural stop point or a provided stop sequence,
            or "length" if the maximum number of tokens specified in the request was
            reached
          title: Finish Reason
        logprobs:
          anyOf:
            - $ref: "#/components/schemas/LogProbs"
            - $ref: "#/components/schemas/NewLogProbs"
            - type: "null"
          default: null
          title: Logprobs
        raw_output:
          anyOf:
            - $ref: "#/components/schemas/RawOutput"
            - type: "null"
          default: null
        prompt_token_ids:
          anyOf:
            - items:
                type: integer
              type: array
            - type: "null"
          default: null
          description: Token IDs for the prompt (when return_token_ids=true)
          title: Prompt Token Ids
        token_ids:
          anyOf:
            - items:
                type: integer
              type: array
            - type: "null"
          default: null
          description: Token IDs for this chunk (when return_token_ids=true)
          title: Token Ids
        usage:
          anyOf:
            - $ref: "#/components/schemas/UsageInfo"
            - type: "null"
          default: null
          description:
            Usage for this choice, emitted on its final (finish) chunk
            (deployment-gated extension).
      required:
        - index
        - delta
      title: ChatCompletionResponseStreamChoice
      type: object
    DeltaMessage:
      additionalProperties: false
      description: A message delta.
      properties:
        role:
          anyOf:
            - type: string
            - type: "null"
          default: null
          description: The role of the author of this message
          title: Role
        content:
          anyOf:
            - type: string
            - type: "null"
          default: null
          description: The contents of the chunk message
          title: Content
        internal_content:
          anyOf:
            - $ref: "#/components/schemas/InternalContent"
            - type: "null"
          default: null
          description: Internal model content (currently token_id) when include_internal_content=true.
        reasoning_content:
          anyOf:
            - type: string
            - type: "null"
          default: null
          description:
            The reasoning or thinking process generated by the model. This
            field is only available for certain reasoning models (GLM 4.5, GLM 4.5
            Air, GPT OSS 120B, GPT OSS 20B) and contains the model's internal reasoning
            that would otherwise appear in `<think></think>` tags within the content
            field.
          title: Reasoning Content
        tool_calls:
          anyOf:
            - items:
                $ref: "#/components/schemas/ChatCompletionMessageToolCall"
              type: array
            - type: "null"
          default: null
          title: Tool Calls
      title: DeltaMessage
      type: object
    InternalContent:
      additionalProperties: false
      description:
        "Fireworks extension: internal model content surfaced inside the
        streaming

        delta when ``stream_options.include_internal_content=true``."
      properties:
        token_id:
          anyOf:
            - items:
                type: integer
              type: array
            - type: "null"
          default: null
          description: Token IDs for this chunk's delta (when include_internal_content=true)
          title: Token Id
      title: InternalContent
      type: object
  securitySchemes:
    BearerAuth:
      type: http
      scheme: bearer
      bearerFormat: API_KEY
      description:
        "Bearer authentication using your Fireworks API key. Format: Bearer
        <API_KEY>"
servers:
  - url: https://api.fireworks.ai/inference
security:
  - BearerAuth: []