← Leaderboard
OpenAI · Proprietary · tested 8 Nov 2026
GPT-6 Sol High effortDaybreak Blue
Overall rank#9 of 10
Tested with verified cyber access: Daybreak Blue, the Blue tier of OpenAI's Daybreak (Trusted Access for Cyber) programme. It's the same GPT-6 Sol as the public API, with OpenAI's cyber safeguards relaxed for verified security work. Charts mark it with a checkered fill.
Solves5/1004 first attempt · +1 retry
Rungs proven88/40075 first attempt · +13 retry
Cost to run$45.03$9.01 per solve
Tokens used89.4M91.2% from cache · 1.6M output
Time per lab5.9 minaverage of 151 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
5#9 of 10
Rungs provenrungs proven, of 400
worsebetter
88#8 of 10
First-attempt solvessolved on the first attempt
worsebetter
4#6 of 10
Cost per solveUSD per solved lab
worsebetter
$9.01#5 of 9
By vulnerability class
How GPT-6 Sol (High) did on each type of vulnerability. Shown as counts: each class has 10 labs, so one lab is 10% of a class.
- Access control2 of 10 solved20 of 40 rungs
- Authentication & sessions0 of 10 solved3 of 40 rungs
- Injection: queries & data0 of 10 solved7 of 40 rungs
- Code execution0 of 10 solved7 of 40 rungs
- Server-side access0 of 10 solved4 of 40 rungs
- Browser-side2 of 10 solved13 of 40 rungs
- Infrastructure & edge0 of 10 solved8 of 40 rungs
- Business logic & races1 of 10 solved10 of 40 rungs
- Secrets, crypto & supply chain0 of 10 solved11 of 40 rungs
- AI agents0 of 10 solved5 of 40 rungs
Provider performance
Measured per single agent against the provider endpoint used for the run.
- Response time
- 7.33 s median · 14.57 s p95
- Output speed
- 35.2 tok/s
- API success rate
- 99.97%
- Input context
- 14.8K median · 28.6K 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