TPSTokens Per Second

Qwen3.8-Max

Alibaba / Qwen🇨🇳 China
active
Compare with other models →
Context window1000K tokens
Input / 1M tokens$2
Output / 1M tokens$6

Version History

3.8-Maxmajor

Qwen3.8-Max scales to 2.4 trillion total parameters (95 billion active) and is built on the Qwen3.5 architecture, with training focused on sustaining autonomous, multi-day agentic tasks rather than single-turn responses. It is the first Qwen-Max model to be open-weighted, with weights arriving on Hugging Face and ModelScope next week.

Benchmark Scores

Full leaderboard →
1497.0 elo
Arena Elo
92.6%
GPQA
41.7%
Hallucination Rate
80.0 tokens_per_sec
Speed (tok/s)

Coverage