Qwen3 32B

Apr 2025
Qwen3 32B
Compare in
on
—per task?
$0.70 / $2.80per 1M in / out
89 tok/sfirst token in 1.3 s?
33Kcontext

Rankingin non-thinking mode?

Strong at

C
Coding#75 of 132Agents & work tasks#122 of 175Tool calling#148 of 242Reasoning#305 of 334Science#246 of 312Knowledge#207 of 318Instructions#205 of 245

Alternatives?

ModelScoreCostPer taskWhat changes
CheaperR1 0528unstatedOpen+79%-43%$0.39 / $1.63Memory (4-bit): 20 GB → 411 GBBetterGPT-5high×2.2+181%$1.25 / $10OpenQwQ 32B-48%-83%$0.15 / $0.40CheaperKimi K2 0711unstatedOpen+48%-29%$0.50 / $2Memory (4-bit): 20 GB → 616 GB+1CheaperGPT OSS 120BhighOpen+4%-95%$0.092Memory (4-bit): 20 GB → 70 GB

Run it yourself

estimate
Fits on✗ 8 GB (phone, laptop)✗ 16 GB✓ 24 GB GPU · compressed✓ 48 GB GPU · 8-bit✓ 80 GB GPU · full quality
LicenceApache-2.0free for commercial use

Benchmarks

all categories
Chat & Knowledge
Humanity's Last Exam4.1#308
AA Intelligence Index 4.3.2≈7.3#268
GPQA Diamond53.5#249
AIME 2025 · AA archived19.7#114
Coding
LiveCodeBench · AA archived28.8#125
Writing & Language
IFBench (instructions)31.5#207

FAQ

How much does Qwen3 32B cost?›

$0.70 per 1M input tokens and $2.80 per 1M output tokens; cheapest via DeepInfra at $0.08 / $0.28.

Is Qwen3 32B free to use?›

There is no free API; the cheapest paid price is $0.08 per 1M input tokens. The weights are open, so you can run it yourself.

How good is Qwen3 32B, and what is it best at?›

#305 of 334 in Chat & Knowledge, #75 of 132 in Coding, #111 of 133 in Agents & Tool Use, #130 of 195 in Writing & Language, #109 of 133 in Vision & Documents.

Is Qwen3 32B better than Qwen2.5 Coder 32B Instruct?›

On Aider Polyglot · two attempts, Qwen3 32B scores 40 vs 16.4 — higher.

How can I use Qwen3 32B?›

Through the API of AWS Bedrock, DeepInfra, SiliconFlow, Tenstorrent.

What is the context window of Qwen3 32B?›

33K tokens, with up to 16K output tokens.

When was Qwen3 32B released? Is there a newer version?›

Released 2025-04-27. It has been superseded by Qwen3 30B A3B 2507 (Reasoning).

Is Qwen3 32B open source?›

Yes, the weights are open (Apache-2.0 licence).

Can I run Qwen3 32B locally?›

The weights are open and can be self-hosted; hardware needs are not confirmed yet.

Can Qwen3 32B read images?›

No, it accepts text only.

What is Qwen3 32B?›

Qwen3 32B is an open-weights language model from Alibaba (Qwen), released Apr 2025, the successor to Qwen2.5 Coder 32B Instruct.

What are the best alternatives to Qwen3 32B?›

R1 0528 (cheaper), GPT-5 (better), QwQ 32B (open weights).