← Leaderboard
xAI · Proprietary · tested 14 Nov 2026
Grok 4.7 High effort
Overall rank#4 of 10
Solves18/10011 first attempt · +7 retry
Rungs proven152/400118 first attempt · +34 retry
Cost to run$748$41.55 per solve
Tokens used1.23B94.9% from cache · 9.6M output
Time per lab16.6 minaverage of 164 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
18#4 of 10
Rungs provenrungs proven, of 400
worsebetter
152#4 of 10
First-attempt solvessolved on the first attempt
worsebetter
11#2 of 10
Cost per solveUSD per solved lab
worsebetter
$41.55#8 of 9
By vulnerability class
How Grok 4.7 (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 solved18 of 40 rungs
- Authentication & sessions0 of 10 solved15 of 40 rungs
- Injection: queries & data1 of 10 solved11 of 40 rungs
- Code execution3 of 10 solved19 of 40 rungs
- Server-side access2 of 10 solved18 of 40 rungs
- Browser-side1 of 10 solved8 of 40 rungs
- Infrastructure & edge1 of 10 solved9 of 40 rungs
- Business logic & races3 of 10 solved18 of 40 rungs
- Secrets, crypto & supply chain4 of 10 solved25 of 40 rungs
- AI agents1 of 10 solved11 of 40 rungs
Provider performance
Measured per single agent against the provider endpoint used for the run.
- Response time
- 6.51 s median · 12.46 s p95
- Output speed
- 79.1 tok/s
- API success rate
- 99.94%
- Input context
- 69.4K median · 146.0K 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