Qwen3.8 README

The project's README from github.com/QwenLM/Qwen3.8. As one markdown file: README.md.

Qwen3.8

💜 Qwen Studio | 🤗 Hugging Face | 🤖 ModelScope | 💬 WeChat (微信) | 🫨 Discord

Welcome to the GitHub repository of the Qwen3.5 open model series, including Qwen3.5, Qwen3.6, and the latest Qwen3.8. Here, you can find official information about Qwen3.8, post your questions (Issues), and share your ideas with the community (Discussions).

Introduction

Qwen3.8

For the first time, Qwen3.8 brings a Qwen-Max-class model to open release. Built on the architectural foundation of Qwen3.5, Qwen3.8 delivers substantial gains across coding, professional work, research, and long-horizon agentic tasks. Beyond answering harder questions, Qwen3.8 is designed to carry complex, multi-step tasks through to completion with greater reliability.

Qwen3.8 features the following enhancements:

Qwen3.6

Building upon the fundamental breakthroughs of Qwen3.5, Qwen3.6 prioritizes stability and real-world utility. It offers developers a more intuitive, responsive, and genuinely productive coding experience, shaped by direct community feedback. This update delivers substantial upgrades, particularly in:

Qwen3.5

Over recent months, we have intensified our focus on developing foundation models that deliver exceptional utility and performance. Qwen3.5 represents a significant leap forward, integrating breakthroughs in multimodal learning, architectural efficiency, reinforcement learning scale, and global accessibility to empower developers and enterprises with unprecedented capability and efficiency.

Qwen3.5 features the following enhancements:

News

Models

The official model weights are released on:

Benchmarks

Qwen3.8 Open Models

For detailed results, please check out the Qwen3.8-2.4T-A95B Model Card and the Qwen3.8-27B Model Card.

Qwen3.6 Open Models

Qwen3.6-27B Benchmark Results

Qwen3.6-35B-A3B Benchmark Results

For detailed results, please check out the Qwen3.6-35B-A3B blog and the Qwen3.6-27B blog.

Qwen3.5 Open Models

Qwen3.5-397B-A17B Benchmark Results

Qwen3.5-122B-A10B, Qwen3.5-35B-A3B, and Qwen3.5-27B Benchmark Results

Qwen3.5-9B and Qwen3.5-4B Benchmark Results

For detailed results, please check out the Qwen3.5 blog.

Quickstart

Official

You can try Qwen3.8 on our official sites and enjoy the native experience with extra features, such as deep research, web dev, and adaptive tool use.

Qwen Studio

If you simply want to try Qwen3.8, Qwen Studio is an AI assistant for everyone. It’s free to use, open to all, and ready to help with creativity, collaboration, and endless possibilities.

Qoder

Qwen3.8 is now available directly on Qoder. Qoder is an agentic coding platform designed for real software development. Qoder is available as a standalone application. Follow its documentation to get started!

QwenWork

Qwen3.8 is now available directly on QwenWork. QwenWork is a one-stop AI working platform launched by Alibaba. Follow its documentation to get started!

Qwen API

QwenCloud provides first-class support for Qwen3.8, which is compatible with various API specifications, including OpenAI and Anthropic, making it simple for you to try Qwen3.8 in your own applications.

Qwen Code

Qwen Code is an open-source AI agent for the terminal, optimized for Qwen models. It helps you understand large codebases, automate tedious work, and ship faster. Follow its documentation to get started!

Local Use

Hugging Face Transformers

transformers acts as the model-definition framework in the current open-weight LLM landscape. It also includes functionalities for LLM inference and training. The addition of serving capabilities in transformers makes it much easier to integrate new models in your development.

To launch a server, simply use the transformers serve command:

transformers serve Qwen/Qwen3.8-27B --port 8000 --continuous-batching

An OpenAI-compatible API will be available at http://localhost:8000/v1. See the Serve CLI guide for more information.

llama.cpp

llama.cpp enables LLM inference with minimal setup and state-of-the-art performance on a wide range of hardware. llama.cpp supports the Qwen3.5 open model series (text & vision). Look for models ending with GGUF on Hugging Face Hub.

MLX (Apple Silicon)

If you are running on Apple Silicon, both mlx-lm (text-only) and mlx-vlm (vision + text) support the Qwen3.5 open model series. Look for models ending with MLX on Hugging Face Hub.

Unsloth

Unsloth contains a local UI to run and train LLMs and diffusion models, including Qwen3.8 and more. See the Qwen3.8 guide for running Qwen3.8 quants with Unsloth.

Deployment

The Qwen3.5 open model series is supported by multiple inference frameworks. Here we demonstrate the usage of SGLang, vLLM, and TokenSpeed.

SGLang

SGLang is a fast serving framework for large language models and vision language models. SGLang can be used to launch a server with an OpenAI-compatible API service.

sglang serve --model-path Qwen/Qwen3.8-27B --port 8000 --tp-size 4 --context-length 262144 --reasoning-parser qwen3 --tool-call-parser qwen3_coder

An OpenAI-compatible API will be available at http://localhost:8000/v1.

Also see SGLang Cookbook on serving Qwen3.8.

vLLM

vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs. vLLM can be used to launch a server with an OpenAI-compatible API service.

vllm serve Qwen/Qwen3.8-27B --port 8000 --tensor-parallel-size 4 --max-model-len 262144 --reasoning-parser qwen3 --enable-auto-tool-choice --tool-call-parser qwen3_coder

An OpenAI-compatible API will be available at http://localhost:8000/v1.

Also see vLLM Recipes on serving Qwen3.8.

TokenSpeed

TokenSpeed is a speed-of-light LLM inference engine. TokenSpeed can be used to launch a server with an OpenAI-compatible API service.

tokenspeed serve Qwen/Qwen3.8-27B --port 8000 --tensor-parallel-size 4 --max-model-len 262144 --reasoning-parser qwen3 --enable-auto-tool-choice --tool-call-parser qwen3_coder

An OpenAI-compatible API will be available at http://localhost:8000/v1.

Also see TokenSpeed Recipes on serving Qwen3.8.

Finetuning

We advise you to use training frameworks, including Unsloth, Swift, Llama-Factory, to finetune your models with SFT, DPO, GRPO, etc.

License Agreement

Please find the license file released with the model weights on Hugging Face Hub or ModelScope.

Citation

If you find our work helpful, feel free to give us a cite.

@misc{qwen3.8,
    title  = {{Qwen3.8-Max}: A New Bar for Coding and Cowork},
    author = {{Qwen Team}},
    year   = {2026},
    month  = {August},
    url    = {https://qwen.ai/blog?id=qwen3.8}
}

@misc{qwen3.6-27b,
    title  = {{Qwen3.6-27B}: Flagship-Level Coding in a {27B} Dense Model},
    author = {{Qwen Team}},
    year   = {2026},
    month  = {April},
    url    = {https://qwen.ai/blog?id=qwen3.6-27b}
}

@misc{qwen3.6-35b-a3b,
    title  = {{Qwen3.6-35B-A3B}: Agentic Coding Power, Now Open to All},
    author = {{Qwen Team}},
    year   = {2026},
    month  = {April},
    url    = {https://qwen.ai/blog?id=qwen3.6-35b-a3b}
}

@misc{qwen3.5,
    title  = {{Qwen3.5}: Towards Native Multimodal Agents},
    author = {{Qwen Team}},
    year   = {2026},
    month  = {February},
    url    = {https://qwen.ai/blog?id=qwen3.5}
}

Contact Us

If you are interested in leaving a message to either our research team or product team, join our Discord or WeChat groups!

Qwen3.8 (QwenLM/Qwen3.8)

Searching Database…

4.3K ·GitHub
Что это такое?Qwen3.8 — новейший открытый релиз большой языковой модели от команды Qwen в Alibaba, часть серии Qwen3.5, куда входят также Qwen3.5 и Qwen3.6; ориентирован на кодинг, профессиональную работу, исследования и долгие агентные задачи.
Это готовое чат-приложение или просто веса моделей, которые я загружаю сам?Этот репозиторий — документация вокруг открытых весов моделей, а не упакованное приложение. Вы скачиваете веса (например, Qwen/Qwen3.8-27B) и запускаете их сами через фреймворк либо используете хостинговые инструменты Alibaba вроде Qwen Studio, Qwen API или Qwen Code.
Как начать пользоваться Qwen3.8?Заберите веса из Hugging Face Hub или ModelScope и раздайте их локально с помощью Hugging Face Transformers (transformers serve), vLLM или SGLang, получив API-эндпоинт, совместимый с OpenAI; либо используйте хостинговые Qwen Studio или Qwen API, ничего не запуская самостоятельно.
Какие инструменты инференса реально умеют запускать Qwen3.8?Модель поддерживают SGLang, vLLM и TokenSpeed для серверного развёртывания, llama.cpp для лёгкого локального инференса и mlx-lm/mlx-vlm на Apple Silicon; дообучение поддерживается через Unsloth, Swift и Llama-Factory.
Он работает на Mac с Apple Silicon или нужна GPU от Nvidia?Mac с Apple Silicon поддерживаются через mlx-lm (только текст) и mlx-vlm (зрение + текст); на другом железе используйте квантованные GGUF-сборки с llama.cpp или полноточные веса с vLLM/SGLang, хотя в README минимальные требования к GPU не указаны.
Лицензия Apache-2.0 в этом репозитории — она же и для самих весов модели?Нет — лицензия Apache-2.0 покрывает сам репозиторий, а веса моделей идут со своим файлом лицензии, размещённым рядом с весами на Hugging Face Hub или ModelScope; его нужно проверить отдельно.
В каких размерах выходит Qwen3.8?Среди выпущенных размеров Qwen3.8-27B и куда более крупная модель Qwen3.8-2.4T-A95B типа mixture-of-experts, обе опубликованы на Hugging Face Hub и ModelScope.
Чем Qwen3.8 отличается от Qwen3.6 и Qwen3.5?Qwen3.8 — первый открытый релиз, достигший уровня Qwen-Max, построен на архитектуре Qwen3.5 с более сильным агентным исполнением; Qwen3.6 сосредоточился на стабильности агентного кодинга и сохранении размышлений; Qwen3.5 ввёл унифицированную vision-language MoE-базу с гибридным вниманием.
Можно дообучить Qwen3.8 самому?Да, команда рекомендует дообучать модели с помощью Unsloth, Swift или Llama-Factory методами вроде SFT, DPO или GRPO.
Qwen3.8 понимает изображения или только текст?Базовая архитектура Qwen3.5 обучалась с early fusion на триллионах мультимодальных токенов, давая понимание изображений вместе с текстом; сборки со зрением доступны через mlx-vlm и GGUF-релизы llama.cpp.
Creator
QwenLM
Size
<0.1 MB
Stars
4.3K
Updated
CommitAug 17
Age
1y
Commits · 90 Days