Skip to content
huggingface.co·

🤖Qwen3.8-Max Adds Vision Input and Non-Thinking Support

Vision input & non-thinking support in Qwen's latest version

TL;DR

Qwen Cloud launches Qwen3.8-Max with vision input, non-thinking support, and a massive context length of 1M tokens by default. Key for teams needing robust multi-step task handling.

Qwen Cloud has released Qwen3.8-Max, an enhanced version of the original model featuring vision input capabilities and non-thinking support. This new release is designed to handle complex tasks more reliably with a default context length of 1 million tokens. Developers will find this particularly useful for professional work, research, and long-horizon agentic tasks. The model's architecture includes fine-grained FP8 quantization, making it highly efficient while maintaining performance parity with the original version.

Qwen3.8-Max Adds Vision Input and Non-Thinking Support — huggingface.co

Key Points

1

Qwen3.8-Max features a context length of up to 1,010,000 tokens natively.

2

The model includes fine-grained FP8 quantization with a block size of 128.

3

Vision input and non-thinking support are now available in Qwen3.8-Max.

4

Evaluation metrics show strong performance on benchmarks like Opus 4.8, Fable 5, and GPT 5.6.

5

Qwen Cloud's API service offers managed, scalable inference without infrastructure maintenance.

Why It Matters

If you're working with complex multi-step tasks in Qwen, the new vision input and non-thinking support in Qwen3.8-Max will significantly improve task reliability and efficiency. Teams using Qwen for professional work or research can now handle longer context lengths without performance degradation.

QwenLLMvision-inputnon-thinking-support

Frequently Asked Questions

Why does this matter?

If you're working with complex multi-step tasks in Qwen, the new vision input and non-thinking support in Qwen3.8-Max will significantly improve task reliability and efficiency. Teams using Qwen for professional work or research can now handle longer context lengths without performance degradation.

What happened?

Qwen Cloud launches Qwen3.8-Max with vision input, non-thinking support, and a massive context length of 1M tokens by default. Key for teams needing robust multi-step task handling.

Comments

Subscribe to join the conversation...

Be the first to comment

Enjoyed this article?

Get it daily. 7am. Free. Reads in 5 minutes.

Join 2,948 builders reading daily.

Also get