Qwen3.5 · 35B A3B
Qwen3.5-35B-A3B is a causal language model with a vision encoder released under the Apache 2.0 license and distributed in Hugging Face Transformers format. It uses a hybrid architecture combining Gated Delta Networks and sparse Mixture-of-Experts, with 35B total parameters and 3B activated, a 2048 hidden dimension, 40 layers, and 256 experts (8 routed + 1 shared active). It supports image-text-to-text tasks, has a native context length of 262,144 tokens (extensible up to 1,010,000 tokens), and according to the model card supports 201 languages and dialects. The official card states compatibility with Hugging Face Transformers, vLLM, SGLang, and KTransformers, and notes a hosted variant (Qwen3.5-Flash) with additional production features.
At a glance
- Parameters
- 35B total · 3B active
- Architecture
- Mixture of experts
- Context length
- 262,144 native
- License
- Apache 2.0
- Disk space
- 24 GB
- Software
- Ollama, MLX, vLLM, SGLang
Will it run on your machine?
Save your machine to see a personalized rating and its reasoning.
Add my hardwareBest for
- Image-text-to-text tasks using the model's vision-language pipeline.
- Multilingual text generation across the languages and dialects the model card states are supported.
- Deployment via inference frameworks the model card lists as compatible (e.g., Hugging Face Transformers, vLLM, SGLang, KTransformers).
- Use in Transformers-format workflows, as the repository provides weights and configuration files in that format.
- Use in managed inference via the official Qwen API service provided by Alibaba Cloud Model Studio, as noted in the model card.
Tradeoffs
- The supplied information does not provide hardware requirements such as memory or GPU specifications.
- No license or compatibility details are stated beyond Apache-2.0 and the listed frameworks.
- Performance figures are not provided in the source data beyond benchmark section headings.
- No testing or validation of the model by the assistant is claimed or should be inferred.
- The model card does not supply deployment instructions or configuration details for the listed frameworks beyond compatibility notes.
Ways to run it
vLLM · Linux
Repository-specific example documented in the publisher model card; confirm dependencies and platform in the source.
vllm serve Qwen/Qwen3.5-35B-A3B --port 8000 --tensor-parallel-size 8 --max-model-len 262144 --reasoning-parser qwen3Transformers · See official guide
Repository-specific example documented in the publisher model card; confirm dependencies and platform in the source.
Official instructionsExplore its uses
Keep exploring
GLM-5.3
An open-weight text model focused on complex coding and long-horizon tasks.
Qwen3.8-27B
A dense vision-language model for coding, reasoning, and agent tasks, with image and video understanding.
Qwen2.5 Coder · 7B
Code generation, repair, and reasoning in a compact open-weight coding model.