benchgap
Agentic · tools

Harvey LAB-AA v1.1 leaderboard

As of 2026-11-18, the highest measured score on Harvey LAB-AA v1.1 is 15.0% by GPT-6.2 Astra. 26 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#ModelScoreSource
1GPT-6.2 Astra15.0%measured
2Muse Wonder Full13.1%measured
3Kimi K3.1 Pro12.5%estimated ± 3.5 pp, medium confidence
4GPT-6.2 Sol11.1%measured
5Muse Wonder10.0%measured
6GPT-6.1 Terra9.8%estimated ± 3.5 pp, medium confidence
7Gemini 4 Carbon9.6%estimated ± 3.5 pp, medium confidence
8Gemini 4 Hydrogen8.9%estimated ± 3.5 pp, medium confidence
9GPT-6 Astra8.6%measured
10Qwen4 Max7.5%estimated ± 3.5 pp, medium confidence
11Gemini 3.7 Flash7.5%estimated ± 4.0 pp, medium confidence
12Claude Opus 57.4%estimated ± 4.0 pp, medium confidence
13GPT-5.57.4%estimated ± 4.0 pp, medium confidence
14Claude Fable 57.4%estimated ± 4.0 pp, low confidence
15Gemini 4 Argon7.4%estimated ± 4.0 pp, low confidence
16GPT-5.6 Sol7.4%estimated ± 4.0 pp, low confidence
17Claude Sonnet 57.4%estimated ± 4.0 pp, low confidence
18Gemini 3.8 Flash7.4%estimated ± 4.0 pp, low confidence
19Claude Opus 4.87.4%estimated ± 4.0 pp, low confidence
20Gemini 3.5 Flash7.4%estimated ± 4.0 pp, low confidence
21Claude Opus 4.77.4%estimated ± 4.0 pp, low confidence
22GLM-5.37.4%estimated ± 4.0 pp, low confidence
23Claude Fable 5.27.2%measured
24Claude Sonnet 5.57.1%estimated ± 3.5 pp, medium confidence
25GPT-6.1 Sol6.9%measured
26Inkling 2 Small6.6%estimated ± 3.5 pp, medium confidence
27Claude Fable 5.16.4%measured
28GPT-6 Luna4.3%estimated ± 3.5 pp, low confidence
29Muse Ember4.3%estimated ± 3.5 pp, low confidence
30Claude Opus 5.54.2%measured
31GPT-6 Sol3.6%measured
32Claude Haiku 5.52.3%estimated ± 3.5 pp, low confidence
33Grok 4.71.6%estimated ± 3.5 pp, low confidence
34MiMo-V2.6-Pro0.8%estimated ± 3.5 pp, low confidence
35Mistral Large 40.0%estimated ± 3.5 pp, low confidence
36Qwen3.8-27B0.0%estimated ± 3.5 pp, low confidence

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General