RecursiveMAS/Mixture-Code-Qwen2.5-Coder-3B

TEXT GENERATIONPricing:Input $0.32 / Cached $0.064 / Output $1.6Concurrent Unit Cost:1Model Size:3.1BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Apr 27, 2026License:mitArchitecture:Transformer0.0K Open Weights Featherless Exclusive Cold

RecursiveMAS/Mixture-Code-Qwen2.5-Coder-3B is a 3.1 billion parameter language model with a 32768 token context length, developed by RecursiveMAS. It functions as a specialized Code Specialist Agent within the RecursiveMAS multi-agent framework, designed for code generation and programming tasks. This model excels in collaborative environments where it exchanges and refines latent states with other domain-specific agents, making it suitable for complex, multi-faceted programming challenges rather than standalone plain-text generation.

Loading preview...

Mixture-Code-Qwen2.5-Coder-3B: A Code Specialist Agent

Mixture-Code-Qwen2.5-Coder-3B is a 3.1 billion parameter model built on the Qwen2.5-Coder-3B base, specifically designed as a Code Specialist Agent within the RecursiveMAS multi-agent framework. Unlike standalone language models, this model is not intended for general plain-text generation but rather for specialized code generation and programming-related tasks within a collaborative system.

Key Capabilities

  • Code Generation: Optimized for producing programming code.
  • Multi-Agent Collaboration: Operates within the RecursiveMAS framework, exchanging and refining latent states with other domain-specialized agents.
  • Mixture-Style Integration: Part of a "Mixture-Style" collaboration, where agents focus on specific domains (like code) and work together to generate final responses.
  • Recursive Processing: Leverages latent-space recursion to scale agent collaboration, allowing for iterative refinement of tasks.

Good For

  • Complex Programming Projects: Ideal for scenarios requiring a dedicated code generation component that can interact with other specialized AI agents.
  • Multi-Agent System Development: Suitable for researchers and developers building and experimenting with multi-agent architectures, particularly those utilizing the RecursiveMAS framework.
  • Specialized Code Assistance: When integrated into a larger system, it can provide highly focused code generation capabilities for specific programming challenges.

Finetunes
1 models