GRIPS — leaderboard

Metric: Accuracy (%). Source: huggingface.co. 27 models tracked.

Top models

#ModelScore
1GPT-5.5 (xHigh)96.4
2GPT-5.5 (High)96.1
3GPT-5.595.9
4Gemini 3.5 Flash (High)94.8
5Claude Opus 4.8 (High)93.7
6GPT-5.5 (Low)93.6
7DeepSeek V4 Pro (Reasoning)92.9
8Claude Sonnet 4.6 (High)92.6
9GLM-5.2 (Reasoning)89.2
10DeepSeek V4 Flash (Reasoning)88.2
11GPT-5.4 Mini (High)85.9
12Mistral Medium 3.5 (High)79.2
13Mistral Medium 3.559
14Claude Haiku 4.552.7
15GPT-5.4 Mini48.5

Interactive version: theaggregate.ai/benchmark?slug=grips · How the rankings work · Data refreshed daily, snapshot 2026-07-22.