RPCBench - Error Group - Underspecified Premise - Conditional Localization: leaderboard
Metric: Localization Accuracy (%). Source: github.com. Saturation forecast: Around 2030. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Sonnet 4.6 | 61.11 |
| 2 | Qwen 3.5 35B A3B | 58.5 |
| 3 | Gemini 3.1 Pro (Preview) | 58 |
| 4 | Qwen 3.5 397B A17B | 57.87 |
| 5 | DeepSeek V4 Pro | 56.35 |
| 6 | GPT-5.5 | 55.96 |
| 7 | Qwen 3.5 122B A10B | 55.9 |
| 8 | Qwen 3.5 Plus | 55 |
| 9 | DeepSeek V4 Flash | 54.17 |
| 10 | Llama 3.1 8B Instruct | 48.46 |
| 11 | Llama 3.1 70B Instruct | 38 |
Interactive version: theaggregate.ai/benchmark?slug=rpcbench-error-group-underspecified-premise-conditional-localization · How It Works · Data refreshed daily, snapshot 2026-09-24.