Sign in

Qwen3-30B-A3B-Instruct-2507

Qwen

Hugging FaceArtificial Analysis

BenchClub cloud

Pass rate
92%
33 of 36 checks
Speed
187.6 tok/s
decode, short prompt
At 8k context
161.3 tok/s
decode
Cloud runs
4
speeds within 1.4%
  • Coding11/1292%
  • Tool use10/1283%
  • Long context12/12100%
Suite
[email protected]
Machine
NVIDIA L40S · 48 GB
Engine
ollama 0.40.14

Headline numbers come only from cloud runs. Runs on members’ own machines are listed below and don’t change them.

Arena score
1384
Among open models
#49of 144
Among all on Arena
#173
Votes
23,312

Measured on Qwen3-30B-A3B-Instruct (2507), the served version, not a local build. LMArena, CC BY 4.0, published Oct 28, 2026.

Score by category

CategoryScore
  • Chinese1432
  • French1418
  • Software & IT1417
  • Coding1416
  • Applied math1411
  • Hard prompts, English1407
  • Spanish1404
  • Hard prompts1398
  • Expert1397
  • Math1394
  • Science1393
  • Medicine & health1392
  • English1391
  • Business & finance1389
  • Legal & government1387
  • German1379
  • Multi-turn1378
  • Long prompts1378
  • Non-English1372
  • Russian1369
  • Instructions1363
  • Excluding ties1358
  • Polish1353
  • Japanese1338
  • Writing & literature1336
  • Media & sports1335
  • Korean1331
  • Creative writing1317

BenchClub runs

Run this model
Run by

PassedFailedNot supported