Atlas

Benchmarks

← All benchmarks

Harvey HLAB - Criteria Pass Rate - Immigration

Professional Work · 2026-05-06

The Immigration practice-area split of Harvey's Legal Agent Benchmark held-out evaluation criteria pass rate. This reports pooled rubric criteria passed within that split and is distinct from the all-pass task-score split.

Top models (higher is better)

ModelScore
Muse Wonder Full94.9
Muse Spark 1.294.1
Muse Wonder94.1
Muse Spark 1.193.2
Grok 4.693.1
Muse Spark 1.392.8
Gemini 4 Argon92.4
Hy4 Preview92.0
Ember-191.3
GLM-5.391.3
Gemini 3.8 Flash90.9
Grok 4.790.9
Claude Opus 590.2
Kimi K390.2
DeepSeek V4.1 Flash 101789.8
Grok 4.589.1
Claude Sonnet 5.588.4
Qwen3.8 27B88.4
GLM-5.288.0
GPT-6.1 Sol88.0
MiMo-V2.6-Flash88.0
Gemini 3.7 Flash87.6
Step 5 Preview87.6
Claude Haiku 5.587.4
Claude Opus 5.587.3
MiMo-V2.6-Pro87.3
Sonnet 586.9
Sonnet 4.686.5
Qwen3.8-Max86.5
Opus 4.885.8
DeepSeek V4.1 Flash85.8
GLM-5.185.5
DeepSeek-V4-Flash-073185.1
GPT-6 Astra84.7
GPT-6 Sol84.7
DeepSeek V4 Pro 081384.4
DeepSeek-V4-Pro84.4
GPT-5.6 Sol84.4
Inkling84.4
Nemotron 3 Ultra 550B A55B (NVFP4)84.4
Loading Atlas data…