> ## Documentation Index
> Fetch the complete documentation index at: https://docs.fireworks.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Match Deployment Shape Versions
> Returns the deployment shape versions compatible with the provided deployment create request. Use this to discover a validated shape before creating a deployment with `deployment_shape` set - shapeless deployments (raw accelerator type/count) skip validated-configuration checks and are far more likely to fail at creation.
Use this endpoint to discover the deployment shape versions compatible with a model before creating a deployment. Pass a [Create Deployment](/api-reference/create-deployment) request in the body — at minimum `createDeploymentRequest.deployment.baseModel` — and the response returns the latest validated shape versions that can serve that model, ready to pass as `deploymentShape`.
Match applies the full server-side compatibility logic for you: PEFT base-model resolution (for LoRA addons and live merge, shapes are matched against the PEFT base model), model-type and parameter-count tiers, embedding vs. non-embedding models, the MULTI\_LORA capability gate when `enableAddons` is set, and hiding of FP4 shapes for full-parameter fine-tunes. [List Deployment Shape Versions](/api-reference/list-deployment-shape-versions) only lists versions of a shape you already know and cannot answer "which shapes work with this model?" — use Match for that.
The account in the URL must be **your own account** — the account that will own the deployment (you must be a member of it, so `accounts/fireworks` fails with a permission error). The model in the request body, however, can live in any account you can deploy from, including publisher accounts like `accounts/fireworks`; public shapes from the publisher's account flow into the results.
## Example: Match shapes for a model
```bash theme={null}
curl -X POST "https://api.fireworks.ai/v1/accounts/YOUR_ACCOUNT_ID/deploymentShapeVersions:match" \
-H "Authorization: Bearer $FIREWORKS_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"createDeploymentRequest": {
"deployment": {
"baseModel": "accounts/fireworks/models/gpt-oss-120b"
}
}
}'
```
To match shapes for a deployment with LoRA addons enabled, include `enableAddons` in the request — only shapes with the MULTI\_LORA capability are returned:
```bash theme={null}
curl -X POST "https://api.fireworks.ai/v1/accounts/YOUR_ACCOUNT_ID/deploymentShapeVersions:match" \
-H "Authorization: Bearer $FIREWORKS_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"createDeploymentRequest": {
"deployment": {
"baseModel": "accounts/fireworks/models/gpt-oss-120b",
"enableAddons": true
}
}
}'
```
Then pass one of the returned shape versions as `deploymentShape` in the [Create Deployment](/api-reference/create-deployment) request. See [What is a deployment shape?](/faq-new/deployment-infrastructure/what-is-a-deployment-shape) for why you should always create deployments from a shape.
## OpenAPI
````yaml post /v1/accounts/{account_id}/deploymentShapeVersions:match
openapi: 3.1.0
info:
title: Gateway REST API
version: 5.10.0
servers:
- url: https://api.fireworks.ai
security:
- BearerAuth: []
tags:
- name: AccountService
- name: DeploymentService
- name: Gateway
- name: ModelService
- name: TrainingService
paths:
/v1/accounts/{account_id}/deploymentShapeVersions:match:
post:
tags:
- Gateway
summary: Match Deployment Shape Versions
description: >-
Returns the deployment shape versions compatible with the provided
deployment create request. Use this to discover a validated shape before
creating a deployment with `deployment_shape` set - shapeless
deployments (raw accelerator type/count) skip validated-configuration
checks and are far more likely to fail at creation.
operationId: Gateway_MatchDeploymentShapeVersions
parameters:
- name: account_id
in: path
required: true
description: The Account Id
schema:
type: string
requestBody:
content:
application/json:
schema:
$ref: >-
#/components/schemas/gatewayGatewayMatchDeploymentShapeVersionsBody
description: The deployment create request to match deployment shape versions for.
required: true
responses:
'200':
description: A successful response.
content:
application/json:
schema:
$ref: >-
#/components/schemas/gatewayListDeploymentShapeVersionsResponse
components:
schemas:
gatewayGatewayMatchDeploymentShapeVersionsBody:
type: object
properties:
createDeploymentRequest:
type: object
description: >-
The deployment create request to match deployment shape versions
for. The full CreateDeploymentRequest schema is accepted here —
including fields like `deployment.enableAddons` that affect shape
compatibility — only the most common fields are listed below. See
the Create Deployment reference page for the complete request
schema.
properties:
deployment:
type: object
description: The properties of the deployment being created.
properties:
baseModel:
type: string
description: The base model the deployment will serve.
enableAddons:
type: boolean
description: >-
If true, the deployment can serve LoRA addons. When set,
only shapes with the MULTI_LORA capability are returned.
required:
- createDeploymentRequest
gatewayListDeploymentShapeVersionsResponse:
type: object
properties:
deploymentShapeVersions:
type: array
items:
$ref: '#/components/schemas/gatewayDeploymentShapeVersion'
type: object
nextPageToken:
type: string
description: >-
A token, which can be sent as `page_token` to retrieve the next
page.
If this field is omitted, there are no subsequent pages.
totalSize:
type: integer
format: int32
description: The total number of deployment shape versions.
gatewayDeploymentShapeVersion:
type: object
properties:
name:
type: string
title: >-
The resource name of the deployment shape version. e.g.
accounts/my-account/deploymentShapes/my-deployment-shape/versions/{version_id}
readOnly: true
createTime:
type: string
format: date-time
description: >-
The creation time of the deployment shape version. Lists will be
ordered by this field.
readOnly: true
snapshot:
$ref: '#/components/schemas/gatewayDeploymentShape'
description: Full snapshot of the Deployment Shape at this version.
readOnly: true
validated:
type: boolean
description: If true, this version has been validated.
public:
type: boolean
description: If true, this version will be publicly readable.
latestValidated:
type: boolean
description: |-
If true, this version is the latest validated version.
Only one version of the shape can be the latest validated version.
readOnly: true
capabilities:
type: array
items:
$ref: '#/components/schemas/DeploymentShapeVersionCapability'
description: The capabilities supported by this deployment shape version.
title: >-
A deployment shape version is a specific version of a deployment shape.
Versions are immutable, only created on updates and deleted when the
deployment shape is deleted.
gatewayDeploymentShape:
type: object
properties:
name:
type: string
title: >-
The resource name of the deployment shape. e.g.
accounts/my-account/deploymentShapes/my-deployment-shape
readOnly: true
displayName:
type: string
description: >-
Human-readable display name of the deployment shape. e.g. "My
Deployment Shape"
Must be fewer than 64 characters long.
description:
type: string
description: >-
The description of the deployment shape. Must be fewer than 1000
characters long.
createTime:
type: string
format: date-time
description: The creation time of the deployment shape.
readOnly: true
updateTime:
type: string
format: date-time
description: The update time for the deployment shape.
readOnly: true
baseModel:
type: string
description: >-
Mutable, but only via UpdateDeploymentShape and only to a compatible
model:
one with the same model_type, parameter_count, and embedding status.
The
derived model_type / parameter_count fields (OUTPUT_ONLY, IMMUTABLE)
are
therefore preserved across a swap. Any incompatible change is
rejected.
title: The base model name. e.g. accounts/fireworks/models/falcon-7b
modelType:
type: string
description: The model type of the base model.
readOnly: true
parameterCount:
type: string
format: int64
description: The parameter count of the base model .
readOnly: true
acceleratorCount:
type: integer
format: int32
description: >-
The number of accelerators used per replica.
If not specified, the default is the estimated minimum required by
the base model.
acceleratorType:
$ref: '#/components/schemas/gatewayAcceleratorType'
description: |-
The type of accelerator to use.
If not specified, the default is NVIDIA_A100_80GB.
precision:
$ref: '#/components/schemas/DeploymentPrecision'
description: The precision with which the model should be served.
disableDeploymentSizeValidation:
type: boolean
description: If true, the deployment size validation is disabled.
enableAddons:
type: boolean
description: >-
If true, LORA addons are enabled for deployments created from this
shape.
Deprecated: set enable_addons on the deployment instead.
draftTokenCount:
type: integer
format: int32
description: |-
The number of candidate tokens to generate per step for speculative
decoding.
Default is the base model's draft_token_count.
draftModel:
type: string
description: >-
The draft model name for speculative decoding. e.g.
accounts/fireworks/models/my-draft-model
If empty, speculative decoding using a draft model is disabled.
Default is the base model's default_draft_model.
Deprecated: set default_draft_model on the base model instead.
ngramSpeculationLength:
type: integer
format: int32
description: >-
The length of previous input sequence to be considered for N-gram
speculation.
disableSpeculativeDecoding:
type: boolean
description: >-
DEPRECATED: This field is a no-op. Speculative decoding is
configured on deployment.
enableSessionAffinity:
type: boolean
description: Whether to apply sticky routing based on `user` field.
numLoraDeviceCached:
type: integer
format: int32
title: How many LORA adapters to keep on GPU side for caching
maxContextLength:
type: integer
format: int32
description: >-
The maximum context length supported by the model (context window).
If set to 0 or not specified, the model's default maximum context
length will be used.
presetType:
$ref: '#/components/schemas/DeploymentShapePresetType'
description: Type of deployment shape for different deployment configurations.
title: >-
A deployment shape is a set of parameters that define the shape of a
deployment.
Deployments are created from a deployment shape.
required:
- baseModel
DeploymentShapeVersionCapability:
type: string
enum:
- CAPABILITY_UNSPECIFIED
- MULTI_LORA
default: CAPABILITY_UNSPECIFIED
description: A capability that a deployment shape version supports.
gatewayAcceleratorType:
type: string
enum:
- ACCELERATOR_TYPE_UNSPECIFIED
- NVIDIA_A100_80GB
- NVIDIA_H100_80GB
- AMD_MI300X_192GB
- NVIDIA_A10G_24GB
- NVIDIA_A100_40GB
- NVIDIA_L4_24GB
- NVIDIA_H200_141GB
- NVIDIA_B200_180GB
- AMD_MI325X_256GB
- AMD_MI350X_288GB
- NVIDIA_B300_288GB
- NVIDIA_GB200
- NVIDIA_GB300
default: ACCELERATOR_TYPE_UNSPECIFIED
DeploymentPrecision:
type: string
enum:
- PRECISION_UNSPECIFIED
- FP16
- FP8
- FP8_MM
- FP8_AR
- FP8_MM_KV_ATTN
- FP8_KV
- FP8_MM_V2
- FP8_V2
- FP8_MM_KV_ATTN_V2
- NF4
- FP4
- BF16
- FP4_BLOCKSCALED_MM
- FP4_MX_MOE
default: PRECISION_UNSPECIFIED
title: >-
- PRECISION_UNSPECIFIED: if left unspecified we will treat this as a
legacy model created before
self serve
DeploymentShapePresetType:
type: string
enum:
- PRESET_TYPE_UNSPECIFIED
- MINIMAL
- FAST
- THROUGHPUT
- FULL_PRECISION
- AGENTIC_CODING
- CHAT
- SUMMARIZATION
- MULTI_LORA
default: PRESET_TYPE_UNSPECIFIED
title: |-
- MINIMAL: Preset for cheapest & most minimal type of deployment
- FAST: Preset for fastest generation & TTFT deployment
- THROUGHPUT: Preset for best throughput deployment
- FULL_PRECISION: Preset for deployment with full precision for training & most accurate numerics
- AGENTIC_CODING: Preset for autonomous code generation and analysis for development workflows
- CHAT: Preset for interactive conversational AI for customer engagement
- SUMMARIZATION: Preset for efficient document and content summarization
- MULTI_LORA: Preset for multi-LORA serving (deployments that can enable LORA addons)
securitySchemes:
BearerAuth:
type: http
scheme: bearer
description: >-
Bearer authentication using your Fireworks API key. Format: Bearer
bearerFormat: API_KEY
````