← Leaderboard
Zhipu AI · Open weights · tested 9 Oct 2026
GLM-5.3 Flash Max effort
Overall rank#13 of 25
Solves47/10043 first attempt · +4 retry
Rungs proven246/400218 first attempt · +28 retry
Cost to run$15.03$0.32 per solve
Tokens used320.5M91.8% from cache · 6.5M output
Time per lab18.9 minaverage of 150 attempts
Where it places
Each line is one metric across all 25 configurations. Right is always better; this model is the large dot.
Solveslabs solved, of 100
worsebetter
47#13 of 25
Rungs provenrungs proven, of 400
worsebetter
246#13 of 25
First-attempt solvessolved on the first attempt
worsebetter
43#10 of 25
Cost per solveUSD per solved lab
worsebetter
$0.32#5 of 15
By vulnerability class
How GLM-5.3 Flash (Max) did on each type of vulnerability. Shown as counts, because some classes have only a few labs.
- Authentication / authorisation bypass16 of 28 solved78 of 112 rungs
- Information disclosure8 of 12 solved34 of 48 rungs
- IDOR4 of 11 solved27 of 44 rungs
- SSRF0 of 10 solved9 of 40 rungs
- Business logic5 of 10 solved28 of 40 rungs
- Key or secret exposure8 of 8 solved32 of 32 rungs
- XSS2 of 6 solved11 of 24 rungs
- Token theft3 of 5 solved17 of 20 rungs
Fewer than 5 labs each: one lab changes these a lot, so read them with care.
- Supply chain0 of 4 solved4 of 16 rungs
- Remote code execution0 of 3 solved1 of 12 rungs
- SQL injection1 of 2 solved4 of 8 rungs
- File inclusion0 of 1 solved1 of 4 rungs
Provider performance
Measured per single agent against the provider endpoint used for the run.
- Response time
- 14.32 s median · 36.63 s p95
- Output speed
- 41.5 tok/s
- API success rate
- 99.02%
- Input context
- 32.4K median · 85.4K 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