Alibaba · Qwen3.8
Qwen3.8-2.4T-A95B
The first Max-class Qwen with downloadable weights: a text-only 2.4T-total / 95B-active MoE under a custom license, for teams that want to self-host the base behind Qwen3.8-Max.
- AA Intelligence
- 40#20 of 142
- Input / output per 1M
- $2.00/ $6.00
- Output speed
- 40tok/s
- Context window
- 262K tokens
Composite indexes
- Artificial Analysis Intelligence Index#20 of 14240
Leader: Claude Opus 5.5 at 58
Coding
- Terminal-Bench 4.0#31 of 8811.1%
Leader: Claude Mythos 5.1 at 60.9%
- Terminal-Bench 2.x#20 of 10382%
Leader: Claude Fable 5.1 at 91.4%
- SciCode#26 of 9454.1%
Leader: Claude Opus 5.5 at 66.9%
Agents and tool use
- GDPval-AA#13 of 1031597
Leader: Claude Opus 5.5 at 1846
Reasoning and knowledge
- GPQA Diamond#12 of 14293.5%
Leader: GPT-6 Astra at 96.1%
- Humanity’s Last Exam#28 of 14742.4%
Leader: Claude Opus 5.5 at 61.4%
- CritPt#16 of 13920%
Leader: GPT-5.6 Sol at 32.3%
Vision and long context
- AA Long Context Reasoning#29 of 13980.3%
Leader: Kimi K3 at 88.7%