[Roadmap]: Reasoning Agent after V0.6

[BabyCNM]

• Benchmark:
  • Simple-Bench
  • ARC-AGI
• Async Client Call: Parallelize LLM calling to speed up searching
• Swarm Agent Implementation
• Efficient Mode: Merging Thinker and Grader
• Batch Norm: Normalizing scores for MCTS
• Multi-round Interaction: Implement interaction with the Reasoning Agent, using the default AssistantAgent system message
• Add RAG compatibility with the ReasoningAgent

Backends:

• Add local DeepSeek-R1 as a supported backend
• Add Gemini-thinking mode as a supported backend
• Add o1 & o3 as a supported backend

Integration:

• Enable tool use for reasoning agent.
• Use reasoning agent together with other agents, in a swarm chat for example.

[Hk669]

#293

added simple bench.

[AgentGenie]

@BabyCNM What is "Add RAG compatibility with the ReasoningAgent" about? I am actually thinking to add reasoning agent to RAG pipeline. One use case is to have an reasoning agent to generate sub-queries based on the initial question.

[giorgossideris]

Hello @BabyCNM @Hk669 @davorrunje can I help somehow with this issue? I would be happy to assist.