Atlas

Benchmarks

← All benchmarks

TaxBench Total pass@1

Professional Work · 2026-05-04

Overall TaxBench pass@1 score, computed across tax knowledge, tax calculations, and data retrieval task groups. Higher scores indicate more first-attempt tax tasks solved correctly.

Top models (higher is better)

ModelScore
GPT-5.5 Pro77.7
GPT-5.4 Pro75.8
GPT-5.575.2
Opus 4.672.8
Gemini 3.1 Pro Preview71.9
Opus 4.767.5
Grok 4.1 Fast67.3
GPT-5.2 Pro67.2
Sonnet 4.663.7
Grok 4.2063.2
Gemini 3.1 Flash (benchmark label)63.0
GPT-5.460.8
Sonnet 4.560.0
Gemini 2.5 Pro57.0
GPT-5.251.6
Loading Atlas data…