OpenAI and Anthropic are reportedly pushing multi-agent RL and new collaboration harnesses
sebkrier · x · 2026-07-26
The quoted post argues that both OpenAI and Anthropic appear to have been doing MARL since the Fable 5 / GPT-5.6 Sol generation.
It says both releases reportedly included:
- multi-agent benchmark scores
- new collaboration harnesses for multi-agent systems
The attached image is a meme framing the field as "Multi Agent Systems" everywhere, which reinforces the idea that multi-agent workflows are becoming a major focus.
More from coding & agent
- Anthropic Engineer Unveils Graph-Based Architecture for Multi-Agent Orchestration — blaizedsouza · 2026-07-26
- Claude Code’s Boris says to delete your system prompts and skills every 6 months — ycombinator · 2026-07-26
- IBM releases a 1-hour course on agentic knowledge graphs and multi-agent orchestration — blaizedsouza · 2026-07-26
- Agent evals are becoming production infrastructure, AWS case study says — krishnan · 2026-07-26
- Anthropic engineer releases a 2-hour workshop on graph engineering for agentic systems — pswider · 2026-07-26
- Grok Build fixes procedural game-model edits faster when the change is explicit — Daniel_Farinax · 2026-07-26