Qwen3.8 Max
Alibaba's flagship in this snapshot. Index 45, and an expensive measured task for that score.
No High score is published. Showing Published.
What this workload costs
List price for one call. Reasoning tokens count as output. A cache hit uses the lab’s cache rate when one is published.
Specs
- API id
- Not published here
- Released
- 2 Sep 2026
- Context
- 984k
- Max output
- —
- Input / 1M
- $2.00
- Output / 1M
- $6.00
- Cache read / 1M
- —
- Knowledge
- —
- Inputs
- Text, Image
- Weights
- Qwen license, not MIT. Read it before you embed the weights.
Round USD list from the 23 Sep 2026 open-weights report, which attributes vendor list prices to Artificial Analysis tracking. A 1 Oct yuan-converted tracker quoted about $1.79 / $5.36. The context is the 984k on the Artificial Analysis leaderboard. The date is the 0902 snapshot name. About 2.4T total, 95B active.
Vision input, from the open-weights report.
Use it when
You want Alibaba's flagship specifically, including vision, and you have already accepted the Qwen license.
Leave it when
Value shopping. The index matches GLM-5.3 and trails MiMo-V2.6-Pro, while the measured task cost is $5.41. That usually means a lot of generated tokens.