What's New

Change to BaseGroupChatManager.select_speaker and support for concurrent agents in GraphFlow

We made a type hint change to the select_speaker method of BaseGroupChatManager to allow for a list of agent names as a return value. This makes it possible to support concurrent agents in GraphFlow, such as in a fan-out-fan-in pattern.  

# Original signature:
async def select_speaker(self, thread: Sequence[BaseAgentEvent | BaseChatMessage]) -> str:
  ...

# New signature:
async def select_speaker(self, thread: Sequence[BaseAgentEvent | BaseChatMessage]) -> List[str] | str:
  ...

Now you can run GraphFlow with concurrent agents as follows:

import asyncio

from autogen_agentchat.agents import AssistantAgent
from autogen_agentchat.conditions import MaxMessageTermination
from autogen_agentchat.teams import DiGraphBuilder, GraphFlow
from autogen_ext.models.openai import OpenAIChatCompletionClient


async def main():
    # Initialize agents with OpenAI model clients.
    model_client = OpenAIChatCompletionClient(model="gpt-4.1-nano")
    agent_a = AssistantAgent("A", model_client=model_client, system_message="You are a helpful assistant.")
    agent_b = AssistantAgent("B", model_client=model_client, system_message="Translate input to Chinese.")
    agent_c = AssistantAgent("C", model_client=model_client, system_message="Translate input to Japanese.")

    # Create a directed graph with fan-out flow A -> (B, C).
    builder = DiGraphBuilder()
    builder.add_node(agent_a).add_node(agent_b).add_node(agent_c)
    builder.add_edge(agent_a, agent_b).add_edge(agent_a, agent_c)
    graph = builder.build()

    # Create a GraphFlow team with the directed graph.
    team = GraphFlow(
        participants=[agent_a, agent_b, agent_c],
        graph=graph,
        termination_condition=MaxMessageTermination(5),
    )

    # Run the team and print the events.
    async for event in team.run_stream(task="Write a short story about a cat."):
        print(event)


asyncio.run(main())

Agent B and C will run concurrently in separate coroutines.

• Enable concurrent execution of agents in GraphFlow by @ekzhu in https://github.com/microsoft/autogen/pull/6545

Callable conditions for GraphFlow edges

Now you can use lambda functions or other callables to specify edge conditions in GraphFlow. This addresses the issue of the keyword substring-based conditions cannot cover all possibilities and leading to "cannot find next agent" bug.

NOTE: callable conditions are currently experimental, and it cannot be serialized with the graph.

• Add callable condition for GraphFlow edges by @ekzhu in https://github.com/microsoft/autogen/pull/6623

New Agent: OpenAIAgent

• Feature: Add OpenAIAgent backed by OpenAI Response API by @jay-thakur in https://github.com/microsoft/autogen/pull/6418

MCP Improvement

• Support the Streamable HTTP transport for MCP by @withsmilo in https://github.com/microsoft/autogen/pull/6615

AssistantAgent Improvement

• Add tool_call_summary_msg_format_fct and test by @ChrisBlaa in https://github.com/microsoft/autogen/pull/6460
• Support multiple workbenches in assistant agent by @bassmang in https://github.com/microsoft/autogen/pull/6529

Code Executors Improvement

• Add option to auto-delete temporary files in LocalCommandLineCodeExecutor by @holtvogt in https://github.com/microsoft/autogen/pull/6556
• Include all output to error output in docker jupyter code executor by @ekzhu in https://github.com/microsoft/autogen/pull/6572

OpenAIChatCompletionClient Improvement

• Default usage statistics for streaming responses by @peterychang in https://github.com/microsoft/autogen/pull/6578
• Add Llama API OAI compatible endpoint support by @WuhanMonkey in https://github.com/microsoft/autogen/pull/6442

OllamaChatCompletionClient Improvement

• Add qwen3 support by @mirpo in https://github.com/microsoft/autogen/pull/6528

AnthropicBedrockChatCompletionClient Improvement

• Allow implicit AWS credential setting for AnthropicBedrockChatCompletionClient by @GeorgeEfstathiadis in https://github.com/microsoft/autogen/pull/6561

MagenticOneGroupChat Improvement

• Use structured output for m1 orchestrator by @ekzhu in https://github.com/microsoft/autogen/pull/6540

Other Changes

• Update website 0.5.7 by @ekzhu in https://github.com/microsoft/autogen/pull/6527
• feat: add qwen3 support by @mirpo in https://github.com/microsoft/autogen/pull/6528
• Fix missing tools in logs by @afzalmushtaque in https://github.com/microsoft/autogen/pull/6532
• Update to stable Microsoft.Extensions.AI release by @stephentoub in https://github.com/microsoft/autogen/pull/6552
• fix: CodeExecutorAgent prompt misuse by @Dormiveglia-elf in https://github.com/microsoft/autogen/pull/6559
• Update README.md by @CakeRepository in https://github.com/microsoft/autogen/pull/6506
• fix:Prevent Async Event Loop from Running Indefinitely by @wfge in https://github.com/microsoft/autogen/pull/6530
• Update state.ipynb, fix a grammar error by @realethanyang in https://github.com/microsoft/autogen/pull/6448
• Add gemini 2.5 fash compatibility by @dmenig in https://github.com/microsoft/autogen/pull/6574
• remove superfluous underline in the docs by @peterychang in https://github.com/microsoft/autogen/pull/6573
• Add/fix windows install instructions by @peterychang in https://github.com/microsoft/autogen/pull/6579
• Add created_at to BaseChatMessage and BaseAgentEvent by @withsmilo in https://github.com/microsoft/autogen/pull/6557
• feat: Add missing Anthropic models (Claude Sonnet 4, Claude Opus 4) by @withsmilo in https://github.com/microsoft/autogen/pull/6585
• Missing UserMessage import by @AlexeyKoltsov in https://github.com/microsoft/autogen/pull/6583
• feat: [draft] update version of azureaiagent by @victordibia in https://github.com/microsoft/autogen/pull/6581
• Add support for specifying the languages to parse from the CodeExecutorAgent response by @Ethan0456 in https://github.com/microsoft/autogen/pull/6592
• feat: bump ags version, minor fixes by @victordibia in https://github.com/microsoft/autogen/pull/6603
• note: note selector_func is not serializable by @bassmang in https://github.com/microsoft/autogen/pull/6609
• Use structured output for m1 orchestrator by @ekzhu in https://github.com/microsoft/autogen/pull/6540
• Parse backtick-enclosed json by @peterychang in https://github.com/microsoft/autogen/pull/6607
• fix typo in the doc distributed-agent-runtime.ipynb by @bhakimiy in https://github.com/microsoft/autogen/pull/6614
• Update version to 0.6.0 by @ekzhu in https://github.com/microsoft/autogen/pull/6624

New Contributors

• @mirpo made their first contribution in https://github.com/microsoft/autogen/pull/6528
• @ChrisBlaa made their first contribution in https://github.com/microsoft/autogen/pull/6460
• @WuhanMonkey made their first contribution in https://github.com/microsoft/autogen/pull/6442
• @afzalmushtaque made their first contribution in https://github.com/microsoft/autogen/pull/6532
• @stephentoub made their first contribution in https://github.com/microsoft/autogen/pull/6552
• @CakeRepository made their first contribution in https://github.com/microsoft/autogen/pull/6506
• @wfge made their first contribution in https://github.com/microsoft/autogen/pull/6530
• @realethanyang made their first contribution in https://github.com/microsoft/autogen/pull/6448
• @GeorgeEfstathiadis made their first contribution in https://github.com/microsoft/autogen/pull/6561
• @dmenig made their first contribution in https://github.com/microsoft/autogen/pull/6574
• @holtvogt made their first contribution in https://github.com/microsoft/autogen/pull/6556
• @AlexeyKoltsov made their first contribution in https://github.com/microsoft/autogen/pull/6583
• @bhakimiy made their first contribution in https://github.com/microsoft/autogen/pull/6614

Full Changelog: https://github.com/microsoft/autogen/compare/python-v0.5.7...python-v0.6.0