modelbenchmark.io

QwQ 32B

Alibaba · qwq-32b

Compare

Qwen reasoning model for deliberate problem solving, math, and coding

Specification

alibaba-cn listing

Context
131K
Max output
8K
Released
2024-12
Knowledge cutoff
2024-04
Retires
2025-06-25
Open weights
yes
Input
text
Output
text

Price

US dollars per million tokens · alibaba-cn list price

Input
$0.287
Output
$0.861

Rate limits

as listed

Requests per minute
300

Quality

4 benchmarks · 2 sources

Composite
41st
percentile of 334 scored models
Rank
198 / 334
±0.08 sd
Evidence
4 × 2
3 or more benchmarks, from 2 or more sources
Effort range
one configuration only
coding17th1/3 bench
reasoning45th2/4 bench
math65th1/7 bench

Every score

one row per source and configuration — nothing averaged away

BenchmarkScoreConfigurationSourceRun
OTIS Mock AIME 2024-202559.2 ±6.3defaultEpoch AI2026-08-28
GPQA diamond65.3 ±2.8defaultEpoch AI2026-08-28
Chess Puzzles5.0 ±2.2defaultEpoch AI2026-08-28
Aider polyglot20.9default · aider diffAider2025-03-06

The leading figure for a benchmark is the median across its configurations, so one heroic high-effort run cannot set the number.

Available from 11 hosts

HostIn $/MOut $/MCache rdCache wrContextOutputRetires
alibaba-cnvendor$0.287$0.861131K8K
deepinfra$0.15$0.40131K131K
nebius$0.15$0.4533K33K
nscale$0.18$0.20
hyperbolic$0.20$0.20131K131K
abacus$0.40$0.4033K33K
sambanova · sambanova$0.50$1.0016K16K2025-06-25
cloudflare$0.66$1.0024K24K
cloudflare-workers-ai · cf$0.66$1.0024K24K
fireworks_ai$0.90$0.90131K131K
together_ai$1.20$1.20131K2025-11-13

More from Alibaba

most-hosted first