Atlas

Benchmarks

← All benchmarks

Immigration Bench - All-Pass

Professional Work · 2026-10-28

Share of Immigration Bench's 262 questions on which every rubric item was met, without web search.

Top models (higher is better)

ModelScore
Claude Fable 5.126.3
Muse Wonder Full25.6
Claude Opus 5.524.4
GPT-6.1 Sol22.5
Grok 4.722.5
Claude Sonnet 5.522.1
GPT-6 Astra22.1
Muse Wonder21.8
Kimi K321.4
Muse Spark 1.221.4
GLM-5.321.0
Gemini 4 Argon20.6
MiMo-V2.6-Pro20.6
DeepSeek V4.1 Flash19.5
Mistral Large 4 Preview18.7
Muse Spark 1.318.3
Grok 4.617.6
DeepSeek V4 Pro 081317.2
Gemini 3.8 Flash16.4
GPT-6 Luna16.4
Claude Haiku 5.516.0
Gemini 3.5 Flash-Lite14.1
MiniMax M312.6
Loading Atlas data…