Qwen4-Max
Qwen4-Max is the first model in Qwen's Qwen4 series: a natively multimodal flagship with a 2.4T-parameter backbone, 71B parameters active per token, text, image, and video input, and a 1M-token context window, built for coding, office productivity, and long-horizon agentic work.
Model
Qwen4-Max is the first model of the Qwen4 series and introduces a new architecture with native multimodality. Its backbone has 2.4T parameters, of which 71B are activated per token, and it accepts text, image, and video input across a 1M-token context window. It can produce up to 384K output tokens, with up to 256K of chain-of-thought when deep thinking is enabled. Qwen positions it for coding, office productivity, and long-horizon agentic work, including directly operating professional tools such as CAD, Blender, Unity, and Godot.
Qwen reports that Qwen4-Max outperforms Qwen3.8-Max while using fewer tokens: it scores 52.69 on Work360 against 47.74 for Qwen3.8-Max with roughly 24–35% fewer tokens, and uses roughly 48–49% fewer tokens on Terminal-Bench 4.0. The model supports function calling and structured outputs. Full details are in Qwen's announcement.
Related resources
Latest updates

Scaling Intelligence: A landmark commission and exhibition

Sandbox resourcing: burst, don't reserve
Why you're paying for capacity you aren't using, and how to pay only for what you use.

Modal becomes a gVisor maintainer
Modal runs upwards of 40 million gVisor containers every day. Here's why we bet on it, what we've built into it, and where we want to take it next.
