Qwen3.8 Max

Qwen3.8 Max is a 2.4-trillion-parameter mixture-of-experts flagship from Alibaba's Qwen line, aimed at coding, professional work, multimodal understanding, and long-horizon agentic workflows. It is the latest in the Qwen line, following Qwen3.7 Flash released 2026-07-15. It has a 1,000,000-token context window and a 131,072-token output limit. Pricing starts at $0.338 per million input tokens and $1.01 per million output tokens, with cached input at $0.0676 per million tokens. Weights are not open. It supports reasoning mode and tool calling, accepts image, pdf, text, and video, and returns text.

Against Qwen3.7 Flash: input $0.0282 to $0.338, output $0.1128 to $1.01, cached input $0.003 to $0.0676.

Range across the hosts that serve it: up to $2.50 per 1M input tokens.

• ·2.4T parameter MoE for complex tasks
• ·1M token context window
• ·131K token output limit
• ·Multimodal input: image, pdf, video
• ·Tool calling and reasoning mode

• Reach for it for long-horizon agentic workflows
• Reach for it for coding and professional work
• Reach for it for multimodal understanding tasks

31 hosts serve this model at the price above · the maker's documentation

• 8Artificial Analysis 评测 Step 5 Preview： Intelligence Index 得 44 分，成本约为同级模型 1/2.82026-09-22
• 8Qwen 发布原生全模态模型 Qwen3.8-Omni-Flash，主打音视频智能体任务交付2026-09-18

2 signals of the archive mention this model, each scored against the published bar. This list is rebuilt from the archive at every build, so it cannot go stale.

The registry keeps the six most recent versions of a line. Older ones stay in the file and can be surfaced without a rebuild.

What is the context window size?

1,000,000 tokens.

Is it open weights?

No, weights are not open.

What input types does it accept?

Image, pdf, text, and video.