AutoResearchExam
Agents get 24 hours on a CPU or GPU machine to improve solutions to open-ended ML research tasks through experiments. The score is the area under the improvement curve on a hidden test set, rewarding both quality and speed.
29 tasks over seven areas, including model training, data curation, AI safety and interpretability. Released September 9th, 2026. The hidden-test metric is meant to catch agents that overfit the visible data.
- Made by
- Bespoke Labs
- Released
- September 2026
- Kind
- Machine learning engineering
- Level
- Research frontier
- Scored
- Checked automatically
- Answers
- Agent task
- Tasks
- All public
- Strongest model tested
- GPT-6 Astra #3 on the capability index
- Last model added
- September 2026
- Scores refreshed
- November 12th, 2026
Claude Opus 5.5 leads with 0.61
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 5.5 0.61
- 2 Claude Fable 5.1 0.60
- 3 GPT-6 Astra 0.60
- 4
Gemini 4 Argon 0.59 - 5 Claude Opus 5 0.58
- 6 Grok 4.7 0.52
- 7 GPT-5.6 Sol 0.51
- 8 Grok 4.6 0.49
- 9 Gemini 3.8 Flash 0.48
- 10 Kimi K3 0.43
- 11
Qwen 3.8 Max 0.42 - 12
Muse Spark 1.3 0.38
Source: Official AutoResearchExam leaderboard
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Claude Opus 5.5 | 0.61 | Sep 2026 | Official leaderboard | |
| 2 | Claude Fable 5.1 | 0.60 | Sep 2026 | Official leaderboard | |
| 3 | GPT-6 Astra | 0.60 | Sep 2026 | Official leaderboard | |
| 4 | Gemini 4 Argon | 0.59 | Sep 2026 | Official leaderboard | |
| 5 | Claude Opus 5 | 0.58 | Jul 2026 | Official leaderboard | |
| 6 | Grok 4.7 | 0.52 | Sep 2026 | Official leaderboard | |
| 7 | GPT-5.6 Sol | 0.51 | Jul 2026 | Official leaderboard | |
| 8 | Grok 4.6 | 0.49 | Aug 2026 | Official leaderboard | |
| 9 | Gemini 3.8 Flash | 0.48 | Sep 2026 | Official leaderboard | |
| 10 | Kimi K3 | 0.43 | Jul 2026 | Official leaderboard | |
| 11 | Qwen 3.8 Max | 0.42 | Aug 2026 | Official leaderboard | |
| 12 | Muse Spark 1.3 | 0.38 | Sep 2026 | Official leaderboard |
2 models have held number one on the official leaderboard since its release in September 2026
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official AutoResearchExam 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 AutoResearchExam 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