​

Playground Categories, New User Roles, Fine-Tuning Improvements, and New Models

​

Playground Categories

​

User Roles: Contributor and Inference

• Contributor: Read and write access to resources without administrative privileges
• Inference: Read-only access with the ability to run inference on deployments

​

Fine-Tuning Improvements

• Stop and Resume Jobs: Stop running fine-tuning jobs and resume them later from where they left off. Available for Supervised Fine-Tuning and Reinforcement Fine-Tuning jobs.
• Clone Jobs: Quickly create new fine-tuning jobs based on existing job configurations using the Clone action.
• Download Output Datasets: Download output datasets from Reinforcement Fine-Tuning jobs, including individual files or bulk download as a ZIP archive.
• Download Rollout Logs: Download rollout logs from Reinforcement Fine-Tuning jobs for offline analysis.

​

✨ New Models

• Gemma 3 12B Instruct is now available in the Model Library
• Gemma 3 4B Instruct is now available in the Model Library
• Qwen3 Omni 30B A3B Instruct is now available in the Model Library

​

📚 Documentation Updates

• Deployment Shapes API: Added List Deployment Shapes and Get Deployment Shape endpoints for querying available deployment shapes
• Evaluator APIs: Added Create Evaluator, Update Evaluator, and helper endpoints for evaluator source code, build logs, and upload validation
• Fine-Tuning APIs: Added Resume DPO Job, Resume Reinforcement Fine-Tuning Step, Execute Reinforcement Fine-Tuning Step, and Get Evaluation Job Log Endpoint
• SDK Examples: Added Python SDK example links for direct routing and supervised fine-tuning workflows

​

Reasoning Guide, Prompt Caching Updates, New Models and CLI Updates

​

Reasoning Guide

• Accessing reasoning_content from thinking/reasoning models
• Controlling reasoning effort with the reasoning_effort parameter
• Streaming with reasoning content
• Interleaved thinking for multi-step tool-calling workflows

​

Prompt Caching Updates

• Cached prompt tokens on serverless now cost 50% less than uncached tokens
• Session affinity routing via the user field or x-session-affinity header for improved cache hit rates
• Prompt optimization techniques for maximizing cache efficiency

​

✨ New Models

• Devstral Small 2 24B Instruct 2512 is now available in the Model Library
• NVIDIA Nemotron Nano 3 30B A3B is now available in the Model Library

​

📚 Documentation Updates

• Reasoning Guide: New documentation for working with reasoning models, including reasoning_content, reasoning_effort, streaming, and interleaved thinking (Reasoning)
• Recommended Models: Updated recommendations to include DeepSeek V3.2 for code generation and Kimi K2 Thinking as a GPT-5 alternative (Recommended Models)
• OpenAI Compatibility: Removed stop sequence documentation as Fireworks is now 1:1 compatible with OpenAI’s behavior (OpenAI Compatibility)
• Evaluator APIs: Added REST API documentation for Evaluator and Evaluation Job CRUD operations (Evals API Reference)
• firectl CLI Reference: Updated with new commands including cancel dpo-job, cancel supervised-fine-tuning-job, set-api-key, redeem-credit-code, and evaluator revision management

​

DeepSeek V3.2 on Serverless, Cached Token Pricing, and New Models

​

☁️ Serverless

• DeepSeek V3.2 is now available on serverless

​

Cached Token Pricing Display

​

Evaluations Dashboard Improvements

• Status column showing evaluator build state (Active, Building, Failed)
• Quick filters to filter evaluators and evaluation jobs by status
• Improved table layout with actions integrated into the status column

​

✨ New Models

• DeepSeek V3.2 is now available in the Model Library
• Ministral 3 14B Instruct 2512 is now available in the Model Library
• Ministral 3 8B Instruct 2512 is now available in the Model Library
• Ministral 3 3B Instruct 2512 is now available in the Model Library
• Mistral Large 3 675B Instruct is now available in the Model Library
• Qwen3-VL-32B-Instruct is now available in the Model Library
• Qwen3-VL-8B-Instruct is now available in the Model Library

​

📚 Documentation Updates

• Reranking Guide: Added documentation for using the /rerank endpoint and /embeddings endpoint with return_logits for reranking, including parallel batching examples (Querying Embeddings Models)

​

Audit Logs, Dataset Download, Weighted Training for Reinforcement Fine-Tuning, and New Model

​

Audit Logs in Web App

• Search and filter logs by status and timeframe
• Detailed view panel for individual log entries
• Easy navigation from the console sidebar under Account settings

​

Dataset Download

• Download individual files from a dataset
• Download all files at once with “Download All”
• Access downloads from the Datasets table in the dashboard

​

Weighted Training for Reinforcement Fine-Tuning

​

✨ New Models

• KAT Coder is now available in the Model Library

​

Evaluator Improvements, Kimi K2 Thinking on Serverless, and New API Endpoints

​

Improved Evaluator Creation Experience

• Fork evaluator templates directly from GitHub repositories
• Browse and preview templates before using them
• Create evaluators with a streamlined save dialog
• View evaluators in a new sortable and paginated table

​

MLOps & Observability Integrations

• Weights & Biases (W&B) integration for experiment tracking during fine-tuning
• MLflow integration for model management and experiment logging

​

✨ New Models

• Kimi K2 Thinking is now available in the Model Library
• KAT Dev 32B is now available in the Model Library
• KAT Dev 72B Exp is now available in the Model Library

​

☁️ Serverless

• Kimi K2 Thinking is now available on serverless

​

📚 New REST API Endpoints

• Create Reinforcement Fine-Tuning Step
• List Reinforcement Fine-Tuning Steps
• Get Reinforcement Fine-Tuning Step
• Delete Reinforcement Fine-Tuning Step
• Scale Deployment
• List Deployment Shape Versions
• Get Deployment Shape Version
• Get Dataset Download Endpoint

​

☀️ Sunsetting Build SDK

​

🚀 Improved RFT Experience

​

Supervised Fine-Tuning

  {
    "messages": [
      {"role": "system", "content": "You are a helpful assistant."},
      {"role": "user", "content": "What is the capital of France?"}, 
      {"role": "assistant", "content": "Paris.", "reasoning_content": "The user is asking about the capital city of France, it should be Paris."}
    ]
  }
  {
    "messages": [
      {"role": "user", "content": "What is 1+1?"},
      {"role": "assistant", "content": "2", "weight": 0, "reasoning_content": "The user is asking about the result of 1+1, the answer is 2."},
      {"role": "user", "content": "Now what is 2+2?"},
      {"role": "assistant", "content": "4", "reasoning_content": "The user is asking about the result of 2+2, the answer should be 4."}
    ]
  }

​

Supervised Fine-Tuning

​

🎨 Vision-Language Model Fine-Tuning

• Qwen 2.5 VL 3B Instruct
• Qwen 2.5 VL 7B Instruct
• Qwen 2.5 VL 32B Instruct
• Qwen 2.5 VL 72B Instruct

• Fine-tune on datasets containing both images and text in JSONL format with base64-encoded images
• Support for up to 64K context length during training
• Built on the same Supervised Fine-tuning V2 infrastructure as text models

​

🔧 Build SDK: Deployment Configuration Application Requirement

• .apply() is now required for on-demand and on-demand-lora deployments
• Serverless deployments do not require .apply() calls
• If you do not call .apply(), you are expected to set up the deployment through the deployment page at https://app.fireworks.ai/dashboard/deployments

• Add llm.apply() after creating LLM instances with deployment_type="on-demand" or deployment_type="on-demand-lora"
• No changes needed for serverless deployments
• See updated documentation for examples and best practices

​

🚀 Bring Your Own Rollout and Reward Development for Reinforcement Learning

​

Supervised Fine-Tuning V2

​

🏗️ Build SDK LLM Deployment Logic Refactor

• The id parameter is now required when deployment_type is "on-demand"
• The base_id parameter is now required when deployment_type is "on-demand-lora"
• The deployment_display_name parameter is now optional and defaults to the filename where the LLM was instantiated

​

🚀 Support for Responses API in Python SDK

​

Support for LinkedIn authentication

• Fireworks uses your LinkedIn primary email address for account identification
• You can switch between different Fireworks accounts by changing your LinkedIn primary email
• See our LinkedIn authentication FAQ for detailed instructions on managing email addresses

​

Support for GitHub authentication

​

🚨 Document Inlining Deprecation

• For image processing: Use Vision Language Models (VLMs) like Qwen2.5-VL 32B Instruct
• For PDF processing: Use dedicated PDF processing libraries combined with text-based LLMs
• For structured extraction: Leverage our structured responses capabilities

​

🎯 Build SDK: Reward-kit integration for evaluator development

• Native reward-kit integration for evaluator development
• Automatic packaging of dependencies from pyproject.toml or requirements.txt
• Local testing capabilities before deployment
• Direct integration with Fireworks datasets and evaluation jobs
• Support for third-party libraries and complex evaluation logic

​

Added new Responses API for advanced conversational workflows and integrations

• Continue conversations across multiple turns using the previous_response_id parameter to maintain context without resending full history
• Stream responses in real time as they are generated for responsive applications
• Control response storage with the store parameter—choose whether responses are retrievable by ID or ephemeral

​

Supervised Fine-Tuning V2

• Supports Qwen 2/2.5/3 series, Phi 4, Gemma 3, the Llama 3 family, Deepseek V2, V3, R1
• Longer context window up to full context length of the supported models
• Multi-turn function calling fine-tuning
• Quantization aware training

​

Reinforcement Fine-Tuning (RFT)

​

Diarization and batch processing support added to audio inference

​

🚀 Easier & faster LoRA fine-tune deployments on Fireworks

firectl create deployment "accounts/fireworks/models/<MODEL_ID of lora model>"

1. Create a deployment using firectl create deployment "accounts/fireworks/models/<MODEL_ID of base model>" --enable-addons
2. Then deploy the addon to the deployment: firectl load-lora <MODEL_ID> --deployment <DEPLOYMENT_ID>