Atlas

Benchmarks

← All benchmarks

Tax Agent Bench — Controversy & Precedence All-Pass

Professional Work · 2026-09-23

All-rubric-item pass rate on the 30 controversy-and-precedence questions.

Top models (higher is better)

ModelScore
Claude Opus 5.530.0
Claude Fable 5.126.7
Claude Opus 523.3
GLM-5.320.0
GPT-5.6 Sol20.0
GPT-5.6 Terra16.7
Grok 4.616.7
Muse Spark 1.316.7
GPT-6 Astra13.3
Grok 4.713.3
Kimi K313.3
MiMo-V2.6-Pro13.3
Sonnet 510.0
GPT-5.510.0
GPT-5.6 Luna10.0
Inkling10.0
DeepSeek V4 Pro 08136.7
Gemini 3.7 Flash6.7
Gemini 3.8 Flash6.7
GPT-6 Luna6.7
GPT-6 Sol6.7
Hy4 Preview6.7
MiniMax M36.7
DeepSeek V4.1 Flash3.3
Mercury 2.53.3
MiMo-V2.6-Flash3.3
Loading Atlas data…