Atlas

Benchmarks

← All benchmarks

CWE-bench v2: Access Control

Cybersecurity · 2026-11-02

Pass@1 on the 29 tasks of Collinear AI's K04 weakness class in CWE-bench v2, rounded to a whole point by the publisher. The ten classes are the publisher's own grouping, not an official taxonomy.

Benchmark source: Cwe Bench

Top models (higher is better)

ModelScoreSource
Claude Opus 5.572.0CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI
Gemini 4 Argon65.0CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI
GPT-6 Astra58.0CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI
Muse Wonder57.0CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI
DeepSeek V4.1 Flash 101752.0CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI
Grok 4.749.0CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI
GLM-5.343.0CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI
Kimi K3.1 Pro43.0CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI
Hy4 Preview41.0CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI
Qwen3.8 Max (0902)41.0CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI
Mistral Large 4 Preview31.0CWE-bench v2 leaderboard (snapshot 2026-11-03) · Collinear AI
Loading Atlas data…