OpenAI releases internal data on how models accelerate its own research
Scobleizer · x · 2026-09-07
An OpenAI researcher (kliu128) released data on how models are accelerating research inside OpenAI, arguing recursive self-improvement may be the biggest driver of AI capabilities in the coming years — but it's invisible outside frontier labs, making transparency urgent. He calls on other AI companies to publish similar data to inform public debate on pacing model development. Simon Willison noted interesting details about how OpenAI researchers use coding agents and speculated a mid-July spike in token spend coincided with the internal tool Astra becoming available to employees.
More from coding & agent
- ScienceClaw Hackathon at MIT Media Lab to build decentralized AI agent collectives for science — ProfBuehlerMIT · 2026-10-23
- What should a native agent harness handle beyond tool calling? — Glittering812 · 2026-10-23
- Running ~100 DeepSeek v4.1 flash agents in parallel for under $20/month — zack_overflow · 2026-10-23
- MCP's stateless 2026-07-28 spec drops session-id — how are stateful tools adapting? — Stock-Pumpkin-8859 · 2026-10-23
- Dev adds local MCP server to Mac screenshot app so Claude designs from your references — Embarrassed-Ebb-740 · 2026-10-23
- Open Duet, a self-hosted open-source Claude Duet alternative, hits 6.1k GitHub stars — matchaman11 · 2026-10-23