Qwen  Models

Qwen3.6-35B-A3B

Qwen3.6-35B-A3B is Qwen’s open MoE vision model: 35B parameters, 3B active, 262K context, Apache 2.0. Low-bit GGUF builds fit in 12 GB.

Qwen3.8-Flash-Next

Qwen3.8-Flash-Next carries 125B parameters with 6B active, plus a 51B n-gram embedding. An experimental preview of the Qwen4 architecture.

Qwen3.8-27B

Qwen3.8-27B is a dense 27.8B multimodal model with a 256K context that fits on a 24 GB GPU at 4-bit. Run it locally in Atomic Chat.

Qwen3.8-2.4T-A95B

Qwen3.8-2.4T-A95B is the downloadable version of Qwen 3.8 Max: a 2.4T MoE with 95B active parameters and a 256K context.

Qwen3.6-27B

Qwen3.6-27B is a dense 27.8B vision-language model with a 262K context and thinking on by default. GGUF builds run from 9.39 GB on a 12 GB machine.

Qwen3-8B

An 8.2B dense LLM from Alibaba’s Qwen3 series with switchable thinking mode, strong reasoning, coding, and 100+ language support.

Qwen2.5-7B-Instruct

A 7.61B instruction-tuned LLM from Alibaba’s Qwen2.5 series with strong coding, math, and multilingual ability across 29+ languages.

Qwen2.5-32B-Instruct

A 32.5B instruction-tuned LLM from Alibaba’s Qwen2.5 series with strong coding, math, and 29+ language support.

Qwen2.5-14B-Instruct

A 14.7B instruction-tuned LLM from Alibaba’s Qwen2.5 series with strong coding, math, and 29+ language support.

Qwen2.5-Coder-7B-Instruct

A 7.6B code-specialized LLM from Alibaba’s Qwen2.5-Coder series, tuned for code generation, reasoning, and fixing.

Qwen2.5-Coder-32B-Instruct

A 32.5B code-specialized LLM from Alibaba’s Qwen2.5-Coder series with open-model state-of-the-art coding ability and 128K context.

Qwen3-235B-A22B

A 235B mixture-of-experts LLM from Alibaba’s Qwen3 series that activates 22B parameters and switches between thinking and non-thinking modes.

Qwen3-30B-A3B-Instruct-2507

A 30.5B-parameter (3.3B active) MoE instruct model from Alibaba’s Qwen3 series with 256K context and strong reasoning, coding, and tool use.

Qwen2.5-72B-Instruct

A 72.7B instruction-tuned LLM from Alibaba’s Qwen2.5 series with strong coding, math and multilingual ability across 29+ languages.

Qwen3-4B-Thinking-2507

A 4B reasoning-focused LLM from Alibaba’s Qwen3 series that always thinks step by step, with a 256K context and strong math, coding, and agentic scores.

WebWorld-8B

WebWorld-8B is a text-only Qwen world model that predicts the next web page state from the current state and an action. Apache 2.0.

Qwen3.6-27B-MTP

Qwen3.6-27B-MTP is Qwen3.6-27B served with its Multi-Token Prediction head: the model drafts several tokens per step for faster local generation.

Qwen3-Coder-30B-A3B-Instruct

Qwen3-Coder-30B-A3B-Instruct is Qwen’s agentic coding MoE: 30.5B parameters, 3.3B active per token, 262K native context. Apache 2.0.

Qwen3-30B-A3B

Qwen3-30B-A3B is a 30.5B MoE model that activates 3.3B params per token, with a switchable thinking mode and 100+ languages.

Qwen3-14B

Qwen3-14B is a 14.8B dense model from the Qwen team with a switchable thinking mode and a 131K YaRN context window. The Q4_K_M build is 9.00 GB.

Qwen3-32B

Qwen3-32B is a 32.8B-parameter Apache 2.0 model from Alibaba’s Qwen team, with a switchable thinking mode and 100+ languages. Q4_K_M fits a 24 GB card.