Qwen3-32B
Qwen- Arena score
- 1340
- Among open models
- #70of 148
- Among all on Arena
- #218
- Votes
- 3,926
Measured on Qwen3-32B, the served version, not a local build. LMArena, CC BY 4.0, published Nov 11, 2026.
Score by category
CategoryScoreOpen rankArena rankVotes
- Applied math1420#43#138282
- Math1399#49#161316
- Science1372#64#199798
- Medicine & health1368#61#192212
- Software & IT1366#71#214843
- Expert1362#60#192236
- English1359#73#2212.2K
- Coding1358#75#227513
- Chinese1357#69#214187
- Hard prompts, English1350#77#229729
- German1341#56#184172
- Hard prompts1334#75#2271.2K
- Multi-turn1331#73#228590
- Long prompts1327#69#221504
- Legal & government1325#79#237285
- Business & finance1318#78#229437
- Non-English1317#72#2221.7K
- Russian1311#73#229412
- Instructions1305#76#234858
- Writing & literature1298#74#2361.1K
- Creative writing1297#69#225616
- Excluding ties1296#72#2202.5K
- Media & sports1279#78#241704
| Model | Params | Score |
|---|---|---|
| 752M | — | |
| 2.0B | — | |
| 4.0B | — | |
| 4.0B | — | |
| 8.2B | — | |
| 14.8B | — | |
| 30.5B | 1317 | |
| 30.5B | 1384 | |
| 32.8B | 1340 | |
| 235B | 1394 | |
| 235B | 1419 |
About as strong, smaller
Compare
gemma-3-12b-itGoogle133412.2B
Discussion
- За считанные минуты локальная ИИ модель моделирует корпус для ESP32C3 #ai #qwen3У Павла! · Oct 27, 202612.3K likes
- Qwen3-TTS family is now open sourced: Voice design, clone, and generationPalmik · Jan 22, 2026744 points
- Orthrus-Qwen3: up to 7.8×tokens/forward on Qwen3, identical output distributionFranckDernoncou · May 15, 2026246 points
- How to Clone Voice Locally in Any Language for Free with Qwen3 TTSUmair Alam Chaudhry · Aug 23, 2026165 likes
- Good results fine tuning a local LLM like Qwen 3:0.6B to categorize questionsdev-experiments · Jun 21, 2026215 points
Built on it
Ways to run it
- GGUF · llama.cpp, Ollama, LM StudioQwen/Qwen3-32B-GGUF57 builds
- MLX · Apple silicon: mlx-lm, LM Studiolmstudio-community/Qwen3-32B-MLX-4bit14 builds
- AWQ · vLLM, SGLangQwen/Qwen3-32B-AWQ16 builds
- GPTQ · vLLM, SGLangJunHowie/Qwen3-32B-GPTQ-Int413 builds
Most downloaded
BenchClub runs
Not runnable yet
Runs open once a Qwen3-32B build is registered.
More from Qwen
See more
Qwen3.8-27B6.9M27.8B

Qwen3.5-27B1.7M27.8B

Qwen3.5-35B-A3B1.4M36.0B

Qwen3.5-122B-A10B775.1K125B
Other 20–40B models
See more
gemma-4-31B-itGoogle6.3M31.3B
Ornith-1.0-35BOrnith1.4M35.1B
NVIDIA-Nemotron-3-Nano-30B-A3B-BF16NVIDIA1.3M31.6B
GLM-4.7-FlashZ.ai1.2M31.2B