AutoRLBench 32 Issue / 18 Solution - Solution: leaderboard

AutoRLBench native 32-issue/18-solution task-set edition. Same published root-cause and repaired-workspace protocols, one run per model/task/level. Issue is an LLM judgment against hidden oracle; solution is the harness rerun metric threshold, with reward hacking failing. Overall counts passes over all50runs, not an average of two task-level rates. The source README announces32issue tasks and24solution-capable tasks, but exactly18 have scored solution runs; do not use24as the reported denominator. Exact task-ID sets are pinned and all rates reproduce the native table to its stated one decimal. Separate from the previously reviewed33/20task-set edition. This board reports Repaired run passes its rerun, 18 tasks (%). Higher is better.

Metric: Repaired run passes its rerun, 18 tasks (%). Source: thinking-space.github.io. Saturation forecast: Around February 2028. 10 models tracked. Updated 14 Nov 2026.

Top models

#ModelSettingScoreOverall rank
1Gemini 4 ArgonMax66.7#12
2GPT-6.1 SolMax50#10
3Kimi K3.1 ProMax50#29
4GPT-6 AstraMax44.4#6
5GLM-5.3Max44.4#68
6Claude Fable 5.1Max38.9#15
7Qwen 3.8 MaxMax33.3#59
8Grok 4.7Max22.2#47
9DeepSeek V4 ProMax16.7#191
10Kimi K3Max11.1#56

No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.

Interactive version: theaggregate.ai/benchmark?slug=autorlbench-32-issue-18-solution-solution · How It Works · Data refreshed daily, snapshot 2026-11-16.