ProactBench - Recovery: leaderboard
Pass rate on the Recovery triggers of ProactBench (2026), turns 8 to 10 after the user signals the task is done, where the assistant must add grounded forward-looking value tied to an earlier detail; a GPT-5.4 judge grades each response Pass, Partial or Fail against a rubric written before the model answers. Percent; higher is better.
Metric: Pass rate (%) on the Recovery triggers (turns 8-10: grounded forward-looking value tied to an earlier detail after the user signals completion) over the 198 curated dialogues of ProactBench (about 624 rubric-scored trigger points: 201 Emergent, 232 Critical, 191 Recovery), each trigger judged Pass, Partial or Fail by a GPT-5.4 judge against a rubric written before the model answers; dialogue histories are fixed from a curation run and the model regenerates only at the triggers, sampled once at temperature 0.7; higher is better. Source: arxiv.org. Saturation forecast: Around February 2028. 16 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | GPT-5.5 | 37.2 | #54 | |
| 2 | Claude Opus 4.7 | 22.5 | #85 | |
| 3 | DeepSeek V4 Flash | 18.3 | #267 | |
| 4 | Qwen 3.5 397B A17B Instruct | 17 | #285 | |
| 5 | O4 Mini (2025-04-16) | 16.2 | #370 | |
| 6 | Qwen 3.5 9B Instruct | 15.4 | #678 | |
| 7 | MiMo-V2.5-Pro | 13.1 | #284 | |
| 8 | Gemini 3.1 Pro (Preview) | 11.5 | #99 | |
| 9 | Gemini 2.5 Pro | Medium | 10.5 | #276 |
| 10 | GPT-4o (2024-11-20) | 8.9 | #859 | |
| 11 | Kimi K2.6 | 7.4 | #192 | |
| 12 | Qwen 2.5 7B Instruct | 6.8 | #2723 | |
| 13 | Gemini 2.5 Flash | 5.2 | #463 | |
| 14 | Qwen 3 1.7B | 4.2 | #5860 | |
| 15 | Llama 4 Maverick | 2.1 | #1110 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=proactbench-recovery · How It Works · Data refreshed daily, snapshot 2026-11-21.