Qwen Cloud

Headlines · 5

1. 4d
   Qwen releases Qwen3.8-2.4T-A95B model weights

   Qwen released the model weights and configuration files for Qwen3.8-2.4T-A95B in Hugging Face Transformers format. The 2.4T-parameter MoE model activates 95B parameters per token and is natively designed for 262,144-token context, with extension up to 1,010,000 tokens. Qwen says the artifacts support vLLM, SGLang, and TokenSpeed, while the separate Qwen3.8-Max adds vision input, non-thinking support, a default 1M context length, and built-in tools.

   Qwen · Hugging Face · 🔥 0
2. 4d
   Qwen releases Qwen3.8-Flash-Next-FP8 open weights

   Qwen published FP8-quantized weights and configuration files for Qwen3.8-Flash-Next in Hugging Face Transformers format. According to Qwen, the fine-grained FP8 quantization uses a block size of 128 and has performance metrics close to those of the original model. The repository also lists compatibility with Transformers, vLLM, SGLang, and TokenSpeed, but the supplied report contains no independent compatibility tests or benchmark results.

   Qwen · Hugging Face · 🔥 0
3. 4d
   Qwen Releases Qwen3.8-27B-FP8 Vision-Language Weights

   Qwen’s Hugging Face repository provides FP8-quantized weights and configuration files for the post-trained Qwen3.8-27B-FP8, a 27B-parameter vision-language model. The artifacts use the Hugging Face Transformers format and are reported as compatible with Transformers, vLLM, SGLang and TokenSpeed. Qwen says fine-grained FP8 quantization with a block size of 128 produces performance nearly identical to the original model.

   Qwen · Hugging Face · 🔥 0
4. 6d
   Qwen Releases Qwen3.8-27B Open-Weight Vision-Language Model

   Qwen released the Qwen3.8-27B model weights and configuration files under the Apache 2.0 license in Hugging Face Transformers format. The 27B dense vision-language model handles images and video, supports flexible thinking controls, and natively offers a 262,144-token context that Qwen says can be extended to 1 million tokens. Its artifacts are compatible with Transformers, vLLM, SGLang, and TokenSpeed, while a hosted Qwen Cloud version is still described as coming soon.

   Qwen · Hugging Face · 🔥 0
5. 6d
   Qwen Releases Qwen3.8-Flash-Next Open-Weight Preview

   Qwen published the post-trained weights and configuration files for Qwen3.8-Flash-Next, a 180B-parameter image-text-to-text model presented as an experimental preview of the architecture intended for Qwen4. The artifacts are listed as compatible with Hugging Face Transformers, vLLM, SGLang and TokenSpeed, while Qwen Cloud offers a managed API; Qwen says its hybrid attention, gated residual and n-gram embedding designs improve long-context efficiency and parameter scaling.

   Qwen · Hugging Face · 🔥 0

Experience and discussion from the community

Share my Qwen Cloud experienceAsk about Qwen Cloud