← Leaderboard
DeepSeek · Open weights · tested 13 Nov 2026 · checkpoint 1017
DeepSeek V4.1 Flash Max effort
Overall rank#3 of 10
Solves21/10017 first attempt · +4 retry
Rungs proven168/400138 first attempt · +30 retry
Cost to run$33.42$1.59 per solve
Tokens used4.66B98.9% from cache · 27.0M output
Time per lab24.7 minaverage of 183 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
21#2 of 10
Rungs provenrungs proven, of 400
worsebetter
168#3 of 10
First-attempt solvessolved on the first attempt
worsebetter
17#1 of 10
Cost per solveUSD per solved lab
worsebetter
$1.59#2 of 9
By vulnerability class
How DeepSeek V4.1 Flash (Max) did on each type of vulnerability. Shown as counts: each class has 10 labs, so one lab is 10% of a class.
- Access control6 of 10 solved31 of 40 rungs
- Authentication & sessions3 of 10 solved23 of 40 rungs
- Injection: queries & data3 of 10 solved20 of 40 rungs
- Code execution0 of 10 solved14 of 40 rungs
- Server-side access1 of 10 solved10 of 40 rungs
- Browser-side1 of 10 solved15 of 40 rungs
- Infrastructure & edge1 of 10 solved13 of 40 rungs
- Business logic & races2 of 10 solved13 of 40 rungs
- Secrets, crypto & supply chain3 of 10 solved17 of 40 rungs
- AI agents1 of 10 solved12 of 40 rungs
Provider performance
Measured per single agent against the provider endpoint used for the run.
- Response time
- 4.33 s median · 11.66 s p95
- Output speed
- 145.5 tok/s
- API success rate
- 99.74%
- Input context
- 123.0K median · 278.5K 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