AutoRLBench 45 Issue / 19 Solution: leaderboard
Thinking-Space AutoRLBench 45-issue / 19-solution edition (commit b3389eac) evaluating autonomous RL research agents across 64 tasks (45 issue diagnosis and 19 solution implementation tasks) in sandboxed repositories; overall pass rate (%).
Metric: Pass rate over 45 issue judgments and 19 solution runs (%). Source: thinking-space.github.io. Saturation forecast: Around 2030. 9 models tracked. Updated 21 Nov 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | DeepSeek V4.1 Flash | Max | 29.7 | #67 |
| 2 | GPT-6.1 Sol | Max | 28.1 | #15 |
| 3 | GLM-5.3 | Max | 28.1 | #72 |
| 4 | GPT-6 Astra | Max | 26.6 | #10 |
| 5 | Qwen 3.8 Max | Max | 23.4 | #63 |
| 6 | Grok 4.7 | Max | 21.9 | #51 |
| 7 | Claude Fable 5.1 | Max | 20.3 | #20 |
| 8 | Kimi K3.1 Pro | Max | 17.2 | #36 |
| 9 | Kimi K3 | Max | 7.8 | #61 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=autorlbench-45-issue-19-solution · How It Works · Data refreshed daily, snapshot 2026-11-21.