NanoGPT-Bench
Coding agents with a fixed H100-hour budget and no internet try to improve on a strong NanoGPT speedrun starting point. The score is the share of later human progress recovered.
At launch the best agent recovered 9.3% of five months of human record progress. Different from the community NanoGPT Speedrun record.
- Made by
- Intology
- Released
- May 2026
- Kind
- AI research tasks
- Level
- Research frontier
- Scored
- Checked automatically
- Answers
- Agent task
- Tasks
- All public
- Strongest model tested
- GPT-5.4 #27 on the capability index
- Last model added
- March 2026
- Scores refreshed
- November 12th, 2026
Claude Opus 4.6 leads with 9.30%
Best score per model from the benchmark's own leaderboard, read on November 12th, 2026. Where a model ran with several settings or harnesses, we keep its best.
- 1 Claude Opus 4.6 max 9.30%
- 2 GPT-5.4 xhigh 8.60%
Source: Official NanoGPT-Bench leaderboard
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Claude Opus 4.6 | 9.30% | max, Autoresearch | Feb 2026 | Official leaderboard |
| 2 | GPT-5.4 | 8.60% | xhigh, Codex | Mar 2026 | Official leaderboard |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official NanoGPT-Bench leaderboard
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official NanoGPT-Bench leaderboard
No lab model card reports it yet
We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark.
More in AI research
- PaperBench In use Agent replicates ICML 2024 papers from scratch. LLM judge grades against author-written hierarchical rubrics.
Qwen 3.8 Max 93% - MLS-Bench In use Agent must invent one algorithmic change in an ML research scaffold whose gain transfers across settings. Scored relative to human baselines. Claude Fable 5.1 50.3%
- PostTrainBench 1.2 In use CLI agent gets a base model, one H100 and a time budget to post-train it for a target benchmark. Scored on normalized gain. Claude Fable 5.1 44.6%
- MLE-bench In use Kaggle-style ML engineering competitions. Agent trains models and is scored against competition leaderboards. GPT-6 Astra 93.8%
- KernelBench In use Replace PyTorch operators with faster custom GPU kernels. Scored on correctness and speedup.
MiniMax-M3 28.8% - PMPP-Hard In use Agents write GPU kernels for 69 tasks that must be correct, and on 48 of them also run within 1.25 times the reference time. The score is the area under the solved-tasks curve over a 10-hour budget, out of 690 task-hours.
Inkling 2 Small 494.3