RPCBench - Conditional Strategy Quality: leaderboard
RPCBench tests proactive critique of flawed recommendation premises, locating the premise problem and suggesting a suitable strategy while remaining faithful to the provided evidence. This series reports RPCBench - Conditional Strategy Quality from 2609.00918v1, S3.T2: Strategy Quality (0-100). Higher is better.
Metric: Strategy Quality (0-100). Source: arxiv.org. Saturation forecast: Around February 2027. 11 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.5 397B A17B | 87.21 | #245 |
| 2 | Qwen 3.5 Plus | 86.91 | #232 |
| 3 | DeepSeek V4 Pro | 85.85 | #188 |
| 4 | Qwen 3.5 122B A10B | 85.33 | #335 |
| 5 | DeepSeek V4 Flash | 84.77 | #259 |
| 6 | Claude Sonnet 4.6 | 84.65 | #159 |
| 7 | Qwen 3.5 35B A3B | 83.46 | #492 |
| 8 | GPT-5.5 | 82.64 | #48 |
| 9 | Gemini 3.1 Pro (Preview) | 80.21 | #93 |
| 10 | Llama 3.1 70B Instruct | 59.26 | #1629 |
| 11 | Llama 3.1 8B Instruct | 51.59 | #4318 |
No result here: #4 Claude Opus 5.5, #5 GPT-6 Astra, #9 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=rpcbench-conditional-strategy-quality · How It Works · Data refreshed daily, snapshot 2026-11-08.