RPCBench - Error Group - Inconsistent Premise - Conditional Localization: leaderboard
Metric: Localization Accuracy (%). Source: github.com. Saturation forecast: Estimated already saturated. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Sonnet 4.6 | 94.93 |
| 2 | Qwen 3.5 Plus | 94.16 |
| 3 | Qwen 3.5 397B A17B | 93.6 |
| 4 | Gemini 3.1 Pro (Preview) | 93.56 |
| 5 | Qwen 3.5 122B A10B | 93.36 |
| 6 | DeepSeek V4 Pro | 92.74 |
| 7 | GPT-5.5 | 92.13 |
| 8 | DeepSeek V4 Flash | 92.04 |
| 9 | Qwen 3.5 35B A3B | 91.1 |
| 10 | Llama 3.1 70B Instruct | 65 |
| 11 | Llama 3.1 8B Instruct | 60.28 |
Interactive version: theaggregate.ai/benchmark?slug=rpcbench-error-group-inconsistent-premise-conditional-localization · How It Works · Data refreshed daily, snapshot 2026-09-24.