Qwen 3.8 Max
AI models review
As of Aug 12, 2026, Qwen 3.8 Max is rated 1490 Elo on LM Arena (6,717 Arena votes).
Is Qwen 3.8 Max worth it in 2026?
Qwen 3.8 Max sits at rank 9 of 20 for general text (Elo 1490, about 6717 votes) but climbs to coding rank 3 with a strong coding Elo of 1670, making it a notably asymmetric model that punches well above its general ranking in code tasks. The Artificial Analysis Coding Index of 71.8 versus an Intelligence Index of 58.1 reinforces that coding is its clear strength. At 2 USD per 1M input and 6 USD per 1M output tokens it sits in a mid-range price tier, and the 1,000,000-token context window is a genuine asset for long-document work. Output speed of 50 tokens per second is moderate.
Developers and teams prioritizing coding quality who also need a very large context window for long codebases or documents, and who can absorb mid-range pricing.
General text and chat quality is only middle of the pack at rank 9 of 20, and the output price of 6 USD per 1M tokens makes high-volume non-coding use relatively expensive for what the text Elo delivers.
ReviewPond's conclusion, drawn only from the verified scores and review themes below. Not a paid or editorial opinion.
Benchmark and price facts
- Vendor
- Alibaba
- License
- proprietary
- LM Arena text
- 1490 Elo (#9 of 20)
- LM Arena coding
- 1670 Elo (#3)
- Price (input)
- $2 / 1M tokens
- Price (output)
- $6 / 1M tokens
- Context window
- 1.000.000 tokens
- AA Intelligence Index
- 58.1 / 100
- AA Coding Index
- 71.8 / 100
- Output speed (AA)
- 50 tokens/s
- Human votes
- 6.717
Elo and votes from LM Arena; price and context from OpenRouter. Real fetched values, attributed, never blended.