Qwen3.6-35B-A3B
A mid-sized, open-weight multimodal mixture-of-experts model from Qwen featuring 35 billion total and 3 billion active parameters.
- Mixture-of-Experts (MoE) architecture with 3B active parameters per forward pass
- Multimodal input support encompassing text, image, and video
- 262,144-token context window
- Apache 2.0 license for open-weight deployment
Identity
Qwen3.6-35B-A3B is a mid-sized, open-weight multimodal mixture-of-experts (MoE) model developed by Qwen, a division of Alibaba (model profile, Signal + Noise wire item). Released on April 16, 2026, the model is distributed under the Apache 2.0 license (model profile). It is characterized by a total parameter count of 35 billion, with 3 billion parameters active per forward pass (model profile, Signal + Noise wire item).
What it is
The model is a multimodal system designed to process and generate content across text, image, and video modalities (model profile). It utilizes a Mixture-of-Experts (MoE) architecture, which allows it to maintain the computational efficiency of a 3-billion parameter model while aiming for the performance levels of larger dense models (Signal + Noise wire item). It supports a context window of 262,144 tokens, enabling the processing of extensive datasets or long-form