Skip to main content

Claude Sonnet 5.5 on BlueBench

Claude Sonnet 5.5 ranks 5th of 26 models on BlueBench, Cotool's benchmark for defensive security work, with an overall score of 77.5. It ranks highest in incident response (83.9, 5th of 26) and lowest in detection engineering (78.1, 12th of 26). It averages $1.01 per task, the 11th lowest cost of the 26 models tested.

Anthropic · Closed weight · Updated Oct 2026
Methodology
BlueBench score
77.5
#5 of 26 models
Cost per task
$1.01
#11 cheapest of 26
Reliability
50.0
#4 of 26 models
Best-ranked track
#5
Incident Response · 83.9
Read the Claude Sonnet 5.5 rundown →Verdict, strengths and weaknesses against similar-cost models, and when to pick something else.

Scores by track

BlueBench weights incident response, threat hunting, and detection engineering at 30% each and malware analysis at 10%.

TrackScoreRankTrack leader
Incident Response83.9#5 of 26GPT-6 Astra · 88.0
Threat Hunting66.9#7 of 26Claude Opus 5.5 · 76.1
Detection Engineering78.1#12 of 26GPT-6 Astra · 87.3
Malware Analysis87.8#3 of 26Claude Opus 5.5 · 89.4

Scores by benchmark

Each BlueBench benchmark is a separate intrusion, real or simulated, with its own tasks and grading.

Models ranked nearby

The models closest to Claude Sonnet 5.5 on the overall leaderboard.

ModelRankScoreCost / task
#379.9$6.76
#479.2$3.64
Claude Sonnet 5.5
#577.5$1.01
#677.2$2.89
#776.7$1.30

Frequently asked questions

How good is Claude Sonnet 5.5 at security operations?
Claude Sonnet 5.5 scores 77.5 overall on BlueBench and ranks 5th of 26 models on incident response, threat hunting, detection engineering, and malware analysis. The top model, Claude Opus 5.5, scores 83.2.
Which security tasks is Claude Sonnet 5.5 best at?
BlueBench track scores for Claude Sonnet 5.5, best rank first. Malware Analysis: 87.8 (3rd of 26); Incident Response: 83.9 (5th of 26); Threat Hunting: 66.9 (7th of 26); Detection Engineering: 78.1 (12th of 26).
How much does Claude Sonnet 5.5 cost per security task?
Claude Sonnet 5.5 averages $1.01 per BlueBench task, including repeated attempts. That is the 11th lowest cost of 26 models. The best-scoring cheaper model is DeepSeek V4.1 Flash at $0.17 per task with an overall score of 71.5.
How was Claude Sonnet 5.5 tested?
Every model runs the same 74 tasks across 6 BlueBench benchmarks, built from real intrusion logs and simulated enterprise environments, using the same harness, prompts, tools, and number of trials. Reports are graded against hidden rubrics and detection rules are re-run against the evidence.

Full leaderboards

AI for the blue team.

Run Cotool's harness in your environment to get real security work done

Book a demo