Create

Sign in to ReadmeX

or

How we handle your data: Privacy Policy

ReadmeX

Your AI reading companion

What would you like to catch up on?

Pick a briefing, or ask your own question.

Your community and people briefings stay personal to you.

Story

Qwen releases FP8 weights for Qwen3.8-2.4T-A95B

AI summary

Qwen introduced Qwen3.8 as an open release in its open-model family, with FP8-quantized weights and configuration files for the post-trained Qwen3.8-2.4T-A95B model in Hugging Face Transformers format. The model has 2.4 trillion total parameters and 95 billion activated parameters. Its artifacts are reported as compatible with vLLM, SGLang, TokenSpeed and other listed frameworks. Qwen says fine-grained FP8 quantization with a block size of 128 produces performance metrics nearly identical to those of the original model, but the supplied report provides no independent measurements.

In-depth insight

The release provides FP8 weights and configuration files for a 2.4-trillion-parameter Qwen open model and identifies compatible inference frameworks. Qwen’s claim of near-identical performance remains unverified by independent measurements.

Qwen/Qwen3.8-2.4T-A95B-FP8 Text Generation • 2.4T • Updated 14 days ago • 10.6k • 277


base_model:

  • Qwen/Qwen3.8-2.4T-A95B library_name: transformers license: other license_name: qwen3.8-max license_link: LICENSE pipeline_tag: text-generation

Qwen3.8-2.4T-A95B-FP8

[!Note] This repository contains FP8-quantized model weights and configuration files for the post-trained model in the Hugging Face Transformers format. These artifacts are compatible with vLLM, SGLang, TokenSpeed, etc. The quantization method is fine-grained fp8 quantization with block size of 128, and its performance metrics are nearly identical to those of the original model.

#Qwen#Qwen3.8-2.4T-A95B-FP8#Hugging Face Transformers#vLLM#SGLang#TokenSpeed

Original · 8 min readRead the original
Group chat

No comments yet. Start the conversation.