Home › Models › Alibaba › Qwen 3.8 Max

Qwen 3.8 Max

Qwen 3.8 Max is the strongest multimodal model in the fleet. It is a 1T model with a 1M context window, up to 131K tokens of output, and text plus image input. Intelligence index is 34. Published scores on record are 95.2% MathVision, 93.5% CharXiv, 93.5% LiveCodeBench, 86.6% Terminal-Bench, and 86.1% OSWorld. Use it for charts, document OCR, GUI computer-use, and hard algorithmic work. It does not take video. Kimi K3 does. DeepSeek still leads it on terminal-agent scores.

Visual mathChart & document OCRGUI agentsPolyglot backendK8s automationMultimodal Vision & GUI Agents

Live telemetry is not published for this route yet.

Parameters1T

Context1M

Max output131K

Intelligence index34

Published benchmarks

Terminal-Bench 2.1 86.6%

LiveCodeBench 93.5%

GPQA Diamond 90%approx

MathVision 95.2%

CharXiv 93.5%