Models

Largest Qwen 3 model with hybrid thinking modes for flexible reasoning control.

235B MoE 128k ctx Open

10 providers

Qwen 3 32B

Mid-size Qwen 3 with strong coding and math capabilities. Open weight.

32B 128k ctx Open

Qwen 3 8B

Compact Qwen 3 for edge and single-GPU deployment. Open weight.

8B 128k ctx Open

Qwen 3 TTS

TTS

Qwen 3 text-to-speech model with voice cloning support.

Qwen 3.5 122B

122B MoE (10B active) 128k ctx Open

Large Qwen 3.5 MoE model with 122B total, 10B active parameters.

Qwen 3.5 35B

35B MoE (3B active) 128k ctx Open

Mid-size Qwen 3.5 MoE model with 35B total, 3B active parameters.

Qwen 3.5 397B

397B MoE (17B active) 128k ctx Open

Largest Qwen 3.5 MoE model with 397B total, 17B active parameters.

Qwen 3.5 72B

Native multimodal Qwen with text, image, and video processing.

72B 128k ctx Open

Qwen 3.5 9B

Compact Qwen 3.5 for single-GPU deployment.

9B 128k ctx Open

Qwen3 Embedding 0.6B

Embedding

Compact Qwen3 embedding for edge and low-resource deployment. Apache 2.0.

0.6B Open

Qwen3 Embedding 4B

Embedding

Mid-size Qwen3 embedding balancing performance and efficiency. Apache 2.0.

4B Open

1 provider

Qwen3 Embedding 8B

Embedding

#1 on MTEB multilingual leaderboard. Best open-source embedding model. Apache 2.0.

8B Open

1 provider

Wan 2.2

Top open-source video model with MoE architecture. Trained on 1.5B videos and 10B images.

Wan 2.5

Previous generation Wan video model with 720p generation 30% faster than 2.2.

5 providers

Wan 2.6

Updated Wan video model with improved quality and speed.

Wan 2.7 Video

Latest Alibaba Wan video model with editing, extending, and reference-to-video capabilities.