Claude Code and Codex Beat Human Speedrun Records, But No Model Invented a New Optimizer
AI Engineer · youtube · 2026-09-27
Prime Intellect research engineer Elie Bakouch (creator of SmolLM) set Claude Code and Codex loose on the community's Optimizer Speedrun — training a GPT-2-level model in the fewest steps. Both agents beat the human record.
Behavior differed sharply: Claude Code stopped every 9-10 hours declaring the record unbeatable and sat idle 1/3 of the time; Codex never quit, wrote far more notes, spawned more sub-agents and burned more tokens. In a six-day run, Kimi was the most token-efficient, and a paper only Claude surfaced led to the best record.
His sobering takeaway: none of the models invented a new optimizer — they combined existing ideas for small gains. He closes with an AlphaEvolve-style discovery loop Prime Intellect is building, arguing this research should happen in the open.
More from coding & agent
- Sydney Codex Community Buildathon set for Nov 14 with Cerebras-backed ultrafast compute track — paw_lean · 2026-11-11
- shadcn's improve hits 10k stars: let top models plan, cheap models execute code fixes — shadcn · 2026-11-11
- OpenAI open-sources Codex Security CLI and SDK for AI-driven vulnerability scanning and fixing — openai · 2026-11-11
- Claude Opus 5.5 runs an architecture search and finds AttnRes-Canon, a new top Transformer design — SonglinYang4 · 2026-11-11
- 5 research habits worth knowing: read papers backwards, reproduce first, wire coding agents via MCP — cneuralnetwork · 2026-11-11
- 'Adam can now do CAE': AI agent showcase in engineering simulation draws buzz — burhop · 2026-11-11