Inference
EN
Alibaba

QwQ-32B

QwQ-32B is a reasoning model from Alibaba. 13 benchmarks count toward its index, in 5 categories.

  • Open weights
  • Reasoning
  • Input: text
  • Released 2025-03
RankedEnough results for an index.
IndexAll categories, out of 100.36.3±9.5Median 51.0
CoverageIndex weight that has results.50%
SpeedOutput tokens per second.—Median 67/s
First tokenSeconds to the first output token.—Median 1.3 s
Input / 1MUS dollars per 1M input tokens.—Median $0.60
Output / 1MUS dollars per 1M output tokens.—Median $2.55
ContextMaximum tokens in one request.131KMedian 262K
EloLMArena rating and rank.1329±4 #234Median 1422

Run locally

32.76B parameters
16-bit
69.8 GB
8-bit
37.5 GB
4-bit
21.9 GB
Weights
Hugging Face (opens in a new tab) ↗

Memory need with a short context. The 4-bit copy fits in 24 GB. Compare local models

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.None counted——
CodingCode writing and repair.None counted——
ReasoningLogic problems and puzzles.1 counted25.6—
MultimodalTasks with images and text.None counted——
KnowledgeFacts and expert knowledge.5 counted38.4206 / 268
MultilingualTasks in many languages.1 counted33.0—
InstructionTasks with strict prompt rules.2 counted28.755 / 63
MathMath problems.4 counted41.9120 / 168

Results

13 counted
BenchmarkThe benchmark name.ResultThe score from the source.PlacePlace among results on this benchmark.
MATH-500 Problem SetMath98.0%4 / 27
American Invitational Mathematics Examination 2024Math79.5%15 / 27
OTIS Mock AIME 2024-2025Math59.2% ±6.3113 / 200
GPQA diamond · Epoch AIKnowledge65.3% ±2.8124 / 217
MMLU-ReduxKnowledge90.0%11 / 19
Instruction-Following EvalInstruction83.9%42 / 70
C-EvalKnowledge88.4%8 / 12
American Invitational Mathematics Examination 2025Math69.5%37 / 53
Artificial Analysis Humanity's Last ExamKnowledge7.4%158 / 222
Graduate-Level Google-Proof Q&AKnowledge65.6%102 / 141
MMMLUKnowledge80.9%11 / 15
Artificial Analysis IFBenchInstruction38.8%110 / 146
Artificial Analysis GPQA DiamondKnowledge59.3%157 / 199
Artificial Analysis Long Context ReasoningReasoning26.7%187 / 217
INCLUDEMultilingual69.7%12 / 12
Berkeley Function Calling Leaderboard v3Agentic66.4%—
LiveCodeBench v5Coding62.7%—

Sources

BenchLMCC BY-NC 4.0Epoch AICC BY

History

5 releases
Index37.5 → 36.3
23 Oct 202612 Nov 2026

Questions

How good is QwQ-32B?
Its index is 36.3 out of 100. That is place 231 of 283 ranked models.
What is QwQ-32B best at?
Math: 41.9 out of 100, place 120 of 168.
What is the context window of QwQ-32B?
131K tokens.
Is QwQ-32B open source?
Yes. Its weights are published.
When was QwQ-32B released?
March 2025.

Badge

QwQ-32B on Inference360: place 231 of 283, index 36.3
HTML
<a href="https://inference360.ai/models/qwq-32b"><img src="https://inference360.ai/badges/qwq-32b.svg" alt="QwQ-32B on Inference360: place 231 of 283, index 36.3" height="20"></a>
Markdown
[![QwQ-32B on Inference360: place 231 of 283, index 36.3](https://inference360.ai/badges/qwq-32b.svg)](https://inference360.ai/models/qwq-32b)

Compare with

Sarvam 105B36.7Ling 2.6 Flash37.0Qwen3 235B-A22B38.1

More from Alibaba

All models
Qwen2.5-Omni 7BUnrankedQwen2.5 VL 72B InstructUnrankedQwen PlusUnrankedQwen2.5-MaxUnrankedQwen2.5-1MUnranked