Anthropic · Proprietary · tested 8 Nov 2026
Claude Sonnet 5.5 High effortCVP Defense Access
Tested with verified cyber access: CVP Defense Access, the Defense Access tier of Anthropic's Cyber Verification Program. It's the same Claude Sonnet 5.5 as the public API, with Anthropic's cyber safeguards relaxed for verified security work. Charts mark it with a checkered fill.
Safety refusals affected this result
Anthropic's cyber safeguards stopped attempts in 20 of 100 labs. Stopped attempts score zero. BB Arena is enrolled in Anthropic's cyber program for verified security researchers, and these refusals happened anyway.
With refusals in more than 10 of 100 labs, its score says more about the safeguards than the model, so it isn't ranked: it's on the refusals board. Its cost, tokens and time are left out of comparisons too. How refusals are scored
On the labs it was never refused
Anthropic's safeguards stopped attempts in 20 of 100 labs. On the other 80, Claude Sonnet 5.5 worked without refusals, so here it's compared like for like with the ranked models that did best on those same 80 labs. These labs aren't a random sample of the benchmark, so this doesn't predict a full score.
- Claude Sonnet 5.5 (High)not ranked1 of 80 solved73 of 320 rungs
- Claude Opus 5 (High)#1 overall18 of 80 solved145 of 320 rungs
- Grok 4.6 (High)#2 overall18 of 80 solved145 of 320 rungs
- DeepSeek V4.1 Flash (Max)#3 overall18 of 80 solved140 of 320 rungs
- Grok 4.7 (High)#4 overall18 of 80 solved133 of 320 rungs
By vulnerability class
How Claude Sonnet 5.5 (High) did on each type of vulnerability. Shown as counts: each class has 10 labs, so one lab is 10% of a class.
- Access control0 of 10 solved16 of 40 rungs
- Authentication & sessionsSafety refusals in 1 of 10 labs0 of 10 solved9 of 40 rungs
- Injection: queries & dataSafety refusals in 2 of 10 labs0 of 10 solved7 of 40 rungs
- Code executionSafety refusals in 2 of 10 labs0 of 10 solved6 of 40 rungs
- Server-side accessSafety refusals in 3 of 10 labs0 of 10 solved7 of 40 rungs
- Browser-sideSafety refusals in 6 of 10 labs0 of 10 solved3 of 40 rungs
- Infrastructure & edgeSafety refusals in 1 of 10 labs0 of 10 solved7 of 40 rungs
- Business logic & races1 of 10 solved9 of 40 rungs
- Secrets, crypto & supply chainSafety refusals in 4 of 10 labs0 of 10 solved4 of 40 rungs
- AI agentsSafety refusals in 1 of 10 labs0 of 10 solved7 of 40 rungs
Provider performance
Measured per single agent against the provider endpoint used for the run.
- Response time
- 4.80 s median · 12.97 s p95
- Output speed
- 101.0 tok/s
- API success rate
- 100.00%
- Input context
- 15.4K median · 36.9K 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
- 20 of 100 labs