← Leaderboard
Anthropic · Proprietary · tested 8 Nov 2026
Claude Opus 5 High effortCVP Defense Access
Overall rank#1 of 10
Tested with verified cyber access: CVP Defense Access, the Defense Access tier of Anthropic's Cyber Verification Program. It's the same Claude Opus 5 as the public API, with Anthropic's cyber safeguards relaxed for verified security work. Charts mark it with a checkered fill.
Solves22/1009 first attempt · +13 retry
Rungs proven178/400127 first attempt · +51 retry
Cost to run$923$41.97 per solve
Tokens used705.3M93.6% from cache · 16.6M output
Time per lab24.3 minaverage of 190 attempts
Where it places
Each line is one metric across all 10 ranked configurations. Right is always better; this model is the large dot.
Solveslabs solved, of 100
worsebetter
22#1 of 10
Rungs provenrungs proven, of 400
worsebetter
178#1 of 10
First-attempt solvessolved on the first attempt
worsebetter
9#5 of 10
Cost per solveUSD per solved lab
worsebetter
$41.97#9 of 9
By vulnerability class
How Claude Opus 5 (High) did on each type of vulnerability. Shown as counts: each class has 10 labs, so one lab is 10% of a class.
- Access control3 of 10 solved26 of 40 rungs
- Authentication & sessions2 of 10 solved22 of 40 rungs
- Injection: queries & data1 of 10 solved13 of 40 rungs
- Code execution2 of 10 solved19 of 40 rungs
- Server-side access1 of 10 solved13 of 40 rungs
- Browser-side2 of 10 solved14 of 40 rungs
- Infrastructure & edge2 of 10 solved13 of 40 rungs
- Business logic & races1 of 10 solved15 of 40 rungs
- Secrets, crypto & supply chain6 of 10 solved28 of 40 rungs
- AI agents2 of 10 solved15 of 40 rungs
Provider performance
Measured per single agent against the provider endpoint used for the run.
- Response time
- 9.05 s median · 34.19 s p95
- Output speed
- 71.2 tok/s
- API success rate
- 99.27%
- Input context
- 35.6K median · 73.9K p95
About this result
Every model runs in the same neutral harness, so results compare the models themselves.
- Harness
- BB Arena neutral harness
- Safety refusals
- None