Atlas

Benchmarks

← All benchmarks

Diligence Stack Agent — Auditability & Process

Professional Work · 2026-07-09

Diligence Stack Agent Auditability and Process is the normalized percentage of available rubric points earned for provenance, checks, tool reliability, and a defensible trail, averaged across evaluated work.

Top models (higher is better)

ModelScore
Claude Fable 586.6
Sonnet 584.6
Opus 4.883.6
Kimi K380.0
Grok 4.573.0
GPT-5.6 Sol67.2
GPT-5.6 Luna65.0
GPT-5.6 Terra63.4
Kimi K2.660.0
GPT-5.555.1
DeepSeek-V4-Pro51.0
Gemini 3.5 Flash50.5
Muse Spark 1.150.0
Grok 4.330.0
Loading Atlas data…