Model vendor
Alibaba Cloud / Qwen Team
The Qwen Team from Alibaba Cloud, developing the Qwen series of large language models including state-of-the-art mixture-of-experts and thinking-enabled models
Updated Aug 11, 2026. Data as of 2026-08-11
Models49
Versions52
Open defaults13
CountryCN
Alibaba Cloud / Qwen Team models
Default-version specifications and scored-version capability values are labeled separately.
| AC | Qwen3.8 Max | 89.3 | 80.0% | 1M | Closed weights | Qwen3.8 Max | |
| AC | Qwen3.7 Max | 78.8 | 100.0% | 1M | Closed weights | Qwen3.7 Max | |
| AC | Qwen3.7-Plus | 73.8 | 100.0% | 1M | Closed weights | Qwen3.7-Plus | |
| AC | Qwen3.6 Plus | 67.5 | 100.0% | 1M | Closed weights | Qwen3.6 Plus | |
| AC | Qwen3.5-397B-A17B | 66.3 | 100.0% | 262.1K | Open weights | Qwen3.5-397B-A17B | |
| AC | Qwen3.6-27B | 60.5 | 80.0% | 262.1K | Open weights | Qwen3.6-27B | |
| AC | Qwen3.5-122B-A10B | 59.2 | 100.0% | 262.1K | Open weights | Qwen3.5-122B-A10B | |
| AC | Qwen3.5-27B | 57.2 | 100.0% | 262.1K | Open weights | Qwen3.5-27B | |
| AC | Qwen3.6-35B-A3B | 55.3 | 80.0% | 262.1K | Open weights | Qwen3.6-35B-A3B | |
| AC | Qwen3.5-35B-A3B | 52.0 | 100.0% | 262.1K | Open weights | Qwen3.5-35B-A3B | |
| AC | Qwen3-235B-A22B-Thinking-2507 | 48.5 | 100.0% | 262.1K | Closed weights | Qwen3-235B-A22B-Thinking-2507 | |
| AC | Qwen3 VL 235B A22B Thinking | 45.8 | 100.0% | 262.1K | Closed weights | Qwen3 VL 235B A22B Thinking | |
| AC | Qwen3 VL 235B A22B Instruct | 43.2 | 100.0% | 262.1K | Closed weights | Qwen3 VL 235B A22B Instruct | |
| AC | Qwen3-235B-A22B-Instruct-2507 | 42.4 | 100.0% | 262.1K | Open weights | Qwen3-235B-A22B-Instruct-2507 | |
| AC | Qwen3.5-9B | 41.4 | 100.0% | 262.1K | Open weights | Qwen3.5-9B | |
| AC | Qwen3-Next-80B-A3B-Thinking | 40.9 | 100.0% | 131.1K | Open weights | Qwen3-Next-80B-A3B-Thinking | |
| AC | Qwen3 Max | 40.7 | 80.0% | 262.1K | Closed weights | Qwen3 Max | |
| AC | Qwen3 VL 32B Thinking | 40.5 | 100.0% | N/A | Closed weights | Qwen3 VL 32B Thinking | |
| AC | Qwen3-Next-80B-A3B-Instruct | 36.6 | 100.0% | 131.1K | Open weights | Qwen3-Next-80B-A3B-Instruct | |
| AC | Qwen3-Coder 480B A35B Instruct | 35.5 | 20.0% | 262.1K | Open weights | Qwen3-Coder 480B A35B Instruct | |
| AC | Qwen3 VL 32B Instruct | 35.2 | 100.0% | N/A | Closed weights | Qwen3 VL 32B Instruct | |
| AC | Qwen3.5-4B | 32.4 | 100.0% | N/A | Closed weights | Qwen3.5-4B | |
| AC | Qwen3 30B A3B | 32.1 | 80.0% | 128K | Closed weights | Qwen3 30B A3B | |
| AC | Qwen3 VL 30B A3B Thinking | 31.8 | 100.0% | 262.1K | Closed weights | Qwen3 VL 30B A3B Thinking | |
| AC | Qwen3 32B | 31.2 | 40.0% | 131.1K | Open weights | Qwen3 32B | |
| AC | Qwen3 VL 30B A3B Instruct | 28.8 | 100.0% | 262.1K | Closed weights | Qwen3 VL 30B A3B Instruct | |
| AC | Qwen3 VL 8B Thinking | 28.5 | 100.0% | 262.1K | Closed weights | Qwen3 VL 8B Thinking | |
| AC | Qwen2-VL-72B-Instruct | 25.5 | 0.0% | N/A | Closed weights | Qwen2-VL-72B-Instruct | |
| AC | QwQ-32B | 24.8 | 80.0% | N/A | Closed weights | QwQ-32B | |
| AC | Qwen3 VL 8B Instruct | 24.8 | 100.0% | 262.1K | Closed weights | Qwen3 VL 8B Instruct | |
| AC | Qwen3 VL 4B Thinking | 23.0 | 100.0% | 262.1K | Closed weights | Qwen3 VL 4B Thinking | |
| AC | Qwen2.5 VL 72B Instruct | 22.7 | 0.0% | 131.1K | Open weights | Qwen2.5 VL 72B Instruct | |
| AC | Qwen2.5 72B Instruct | 21.4 | 100.0% | 131.1K | Closed weights | Qwen2.5 72B Instruct | |
| AC | Qwen3 VL 4B Instruct | 19.3 | 100.0% | 262.1K | Closed weights | Qwen3 VL 4B Instruct | |
| AC | Qwen2.5 VL 32B Instruct | 17.9 | 100.0% | N/A | Closed weights | Qwen2.5 VL 32B Instruct | |
| AC | QvQ-72B-Preview | 14.9 | 0.0% | N/A | Closed weights | QvQ-72B-Preview | |
| AC | Qwen2.5 32B Instruct | 14.8 | 80.0% | N/A | Closed weights | Qwen2.5 32B Instruct | |
| AC | Qwen2.5 VL 7B Instruct | 12.8 | 20.0% | N/A | Closed weights | Qwen2.5 VL 7B Instruct | |
| AC | Qwen2.5-Omni-7B | 10.2 | 100.0% | N/A | Closed weights | Qwen2.5-Omni-7B | |
| AC | Qwen2 72B Instruct | 9.8 | 80.0% | N/A | Closed weights | Qwen2 72B Instruct | |
| AC | Qwen2.5 14B Instruct | 8.2 | 80.0% | N/A | Closed weights | Qwen2.5 14B Instruct | |
| AC | Qwen2.5 7B Instruct | 6.5 | 100.0% | 131.1K | Closed weights | Qwen2.5 7B Instruct | |
| AC | Qwen2.5-Coder 32B Instruct | 4.4 | 80.0% | 128K | Closed weights | Qwen2.5-Coder 32B Instruct | |
| AC | Qwen3.5-2B | 2.8 | 60.0% | N/A | Closed weights | Qwen3.5-2B | |
| AC | Qwen2 7B Instruct | 0.0 | 100.0% | N/A | Closed weights | Qwen2 7B Instruct | |
| AC | Qwen2.5-Coder 7B Instruct | 0.0 | 80.0% | N/A | Closed weights | Qwen2.5-Coder 7B Instruct | |
| AC | Qwen3.5-0.8B | 0.0 | 60.0% | N/A | Closed weights | Qwen3.5-0.8B | |
| AC | Qwen Image | N/A | N/A | N/A | Closed weights | N/A | Not scored |
| AC | Qwen3-Coder | N/A | N/A | 256K | Closed weights | Not scored |
Alibaba Cloud / Qwen Team capability scores
Each unique model uses its latest scored version.
Alibaba Cloud / Qwen Team Highlights
Leading Alibaba Cloud / Qwen Team models by score, price and context window.
Top scoreQwen3.8 Max89.3 LLMBoardLowest official inputQwen2.5 Omni 7B$0.10 / 1MLargest contextQwen3.6 Plus1M tokens
Open-weight defaults13of 49 models
What is Alibaba Cloud / Qwen Team?
Background and model catalog information for Alibaba Cloud / Qwen Team.
The Qwen Team from Alibaba Cloud, developing the Qwen series of large language models including state-of-the-art mixture-of-experts and thinking-enabled models
Model ownership is separate from API hosting. One model can be offered by multiple API providers with different prices.
- Vendor
- Alibaba Cloud / Qwen Team
- Models
- 49
- Versions
- 52
- Country
- CN
- Founded
- Not available
Catalog data as of 2026-08-11.
FAQ
Common questions about Alibaba Cloud / Qwen Team.
What is Alibaba Cloud / Qwen Team's highest-ranked model?
Qwen3.8 Max currently has the strongest LLMBoard score from this vendor.
How many Alibaba Cloud / Qwen Team models are tracked?
The catalog contains 49 independently selectable Alibaba Cloud / Qwen Team models and 52 versions.
Does Alibaba Cloud / Qwen Team host every listed model API?
Not necessarily. Vendor identifies the model developer; API providers are tracked separately.
Does Alibaba Cloud / Qwen Team publish open-weight models?
13 current default versions are marked as open weight.