🔙 Go back to more models
Qwen2.5 Series
QwQ-32B
Qwen3 Series
Qwen spans both LLM (language-only) and multimodal (text, vision, audio, etc.) use cases. Qwen includes models for reasoning, coding, agents, vision, and more.
AI Generator
Qwen
AI model for text, coding, images
Qwen is a multi-faceted AI model family developed by Alibaba Cloud, encompassing large language models (LLMs), vision-language models (VL), and more.
- Qwen2.5: Instruction-tuned LLMs in various sizes (e.g., 72B).
- Qwen2.5-Coder: Specialized for coding tasks.
- Qwen2.5-Math: Tailored for advanced mathematical reasoning.
- Qwen2.5-VL: Vision-Language model capable of document understanding and long-video comprehension. Available in multiple sizes up to 72B.
- Qwen2.5-Omni: Multimodal model handling text, image, video, and audio.
- A reasoning-optimized model with 32B parameters. Integrated into Qwen Chat and designed for strong problem-solving capabilities.
- Latest flagship. Includes dense and Mixture-of-Experts (MoE) architectures ranging from 0.6B to 235B parameters, with 22B active parameters in the largest variant.
- Introduces thinking mode (for deep reasoning) and non-thinking mode (for speed), along with a thinking budget mechanism.
- Qwen-VL / Qwen-VL-Chat: Vision-language models focused on understanding and generating image-grounded text, captions, and visual grounding tasks.
- Additional specialized models like Qwen-VL2 (2B & 7B), Qwen-Audio, Qwen2-Math, and Qwen Image (image generation) have been released as well.
