- Qwen3.8-27B has 27 billion parameters and supports image, text and video inputs, according to its model materials. [1] - The model has a native context window of 262,144 tokens and is distributed under Apache 2.0. [1] - Alibaba positions it above the prior Qwen3.7-Plus model for coding and productivity tasks, but the materials reviewed do not establish a like-for-like comparison with competing models.
[1][2] Alibaba’s Qwen team has released Qwen3.8-27B, an open-weight dense multimodal model that supports a native 262,144-token context window. The model is available through Qwen’s Hugging Face and ModelScope releases under the Apache 2.0 license. [1]
The model card describes support for image, text and video inputs, with a focus on coding, agentic applications and office workflows. Alibaba says the 27-billion-parameter model improves on Qwen3.7-Plus in those tasks. [1]
Benchmark claims remain limited #
Alibaba’s positioning is not yet a substitute for a documented comparison. The release materials reviewed do not provide a complete, like-for-like benchmark table covering Qwen3.8-27B, prior Qwen models and competing open models under the same prompts, harnesses and evaluation dates. Claims of superiority therefore remain attributed company claims rather than independently established results. [1][2]
Companies mentioned #
Further sources #
The stories that matter, in one email. Free — unsubscribe anytime.