Atlas

Benchmarks

← All benchmarks

TaxBench Tax Knowledge pass@1

Professional Work · 2026-05-04

Tax Knowledge subset of TaxBench reporting pass@1 first-attempt performance on professional tax tasks.

Top models (higher is better)

ModelScore
GPT-5.5 Pro84.2
Grok 4.1 Fast84.2
GPT-5.4 Pro83.3
GPT-5.2 Pro82.5
Grok 4.2081.7
Gemini 3.1 Flash (benchmark label)80.8
GPT-5.579.3
Opus 4.679.2
Gemini 3.1 Pro Preview78.3
Gemini 2.5 Pro72.5
Sonnet 4.670.8
GPT-5.470.0
Opus 4.765.8
Sonnet 4.564.3
GPT-5.262.5
Loading Atlas data…