RPCBench - Error Group - Underspecified Premise - Proactive Detection: leaderboard

Metric: Detection Rate (%). Source: github.com. Saturation forecast: Around 2033. 11 models tracked.

Top models

#ModelScore
1DeepSeek V4 Pro15.15
2Claude Sonnet 4.611.77
3Qwen 3.5 35B A3B11.77
4Gemini 3.1 Pro (Preview)11.54
5DeepSeek V4 Flash11.08
6Qwen 3.5 122B A10B11.08
7Qwen 3.5 Plus10.77
8Qwen 3.5 397B A17B9.77
9GPT-5.58.38
10Llama 3.1 8B Instruct5
11Llama 3.1 70B Instruct3.85

Interactive version: theaggregate.ai/benchmark?slug=rpcbench-error-group-underspecified-premise-proactive-detection · How It Works · Data refreshed daily, snapshot 2026-09-24.