NanoGPT-Bench: leaderboard
Autonomous coding and ML-research benchmark based on the NanoGPT speedrun, measuring how much historical human optimization progress an agent can recover.
Metric: % of Human Progress Recovered. Source: github.com. Source: www.intology.ai. 3 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.4 (xHigh) | 8.6 | #79 (GPT-5.4) |
| 2 | Claude Opus 4.6 (Max) | 8.2 | #63 (Claude Opus 4.6) |
Interactive version: theaggregate.ai/benchmark?slug=nanogpt-bench · How It Works · Data refreshed daily, snapshot 2026-10-20.