OpenAI's Next Model 'Astra' Leaked: Long-term Memory and Agent Coordination
MattGarciaEth · x · 2026-08-30
Alex Heath previewed OpenAI's next big model, Astra. It reportedly can run for days or weeks, remember corrections, coordinate multiple agents, and navigate desktop software faster than a human. Sam Altman expects Astra to be the first model to invent meaningful new things, with OpenAI aiming for a version that runs forever within ChatGPT and the API.
Related event: Rumors mount around OpenAI's GPT-Astra with early-September launch tipped(18 posts)→
More from AGI Musings
- Geoffrey Irving: stopping to explore now would be a spectacular golden age — geoffreyirving · 2026-10-15
- Ex-UK AISI chief scientist puts odds of superintelligence killing everyone at a coin flip — geoffreyirving · 2026-10-15
- There is no posttraining data commons: open models' next data bottleneck — dlwh · 2026-10-15
- tszzl: agentic auto-research makes mechinterp finally reveal what's inside models' minds — tszzl · 2026-10-15
- Liberal-arts staffer uses AI to reverse-engineer a multiplayer web CS in 20 minutes — FuSheng_0306 · 2026-10-15
- Full-stack data engineer on thriving in AI era by never fixing his identity — teodorio · 2026-10-15