log inget an API key
Platform benchmark

qwen3.6-35b-a3b

qwen · MATH-500 · seed 42 · 2026-11-12 07:18 UTC

ResultsMATH-500 · n=96 · seed 42
57.3%

57.3% correct · 58.3% answered on MATH-500

A low answered rate means the run starved on timeouts — not that the model is weak.

Score per $
2.18
Run cost
$0.26
Latency p50
50.6 s
Latency p95
117.0 s
Items
96/96 completed
Dropped
0

Dataset: MATH-500 · per-item question and response text available in the results.