OpenAI Codex lead Tibo: ChatGPT and Codex will merge into one personal AGI
量子位 · wechat · 2026-08-25
OpenAI Codex lead Thibault Sottiaux ("Tibo", jokingly revered as a godfather for resetting usage limits) sat down with Matthew Berman for a wide-ranging interview covering his Google/DeepMind years and OpenAI's current direction.
Key points: ChatGPT and Codex will ultimately merge into a single highly personalized AGI whose interface adapts to each user's tasks and habits—no separate "coding agent" vs "chat assistant"; the next-gen agent isn't about adding more Skills, Memory or sub-agents but making those mechanisms disappear into a companion that continuously understands you and your team; laptops will become the bottleneck, pushing agents to the cloud; once UltraFast compresses responses to near-human thinking speed, today's workflow of running 10-15 concurrent agents to compensate for slow models will return to real-time flow.
He also said OpenAI is already practicing recursive self-improvement—having its strongest models optimize CUDA kernels, inference stack and infrastructure into a flywheel—and that its differentiation vs Anthropic is "giving the strongest capabilities to as many people as possible," not raw model superiority. He recalled DeepMind building LMChat a year before ChatGPT but being organizationally unable to ship products.
More from coding & agent
- gitea-mcp: MCP Server with 186 Tools for Gitea API Integration — modelcontextprotocol · 2026-08-25
- Long-Horizon Agent Dev Pain: Not Enough Time to Run Full Rollout — agihouse_org · 2026-08-25
- AGI House Hackathon Recap: Long-horizon Agents Need Verification, Not Just Memory — agihouse_org · 2026-08-25
- Terminal-Bench 3.0: Top Model Score Plummets to 43.5% as Agents Fail to Fix Root Causes — ajratner · 2026-08-25
- Spine-Branch Framework Boosts Multi-Agent Success by 16.5% — ZhiyuChen4 · 2026-08-25
- Anthropic Showcases Creative Claude Code Projects: From Medical Viewers to 3D Streets — 机器之心 · 2026-08-25