Qwen 3.8 Max vs Qwen 3.8 Omni Flash
Lower and Larger mark one number at a time; they don’t rank the models. How well a model does depends on your task; see Benchmarks for measured results.
| Detail | Qwen 3.8 Maxalibaba/qwen3.8-max | Qwen 3.8 Omni Flashalibaba/qwen3.8-omni-flash |
|---|---|---|
| Maker | alibaba | alibaba |
| Kind | Chat | Chat |
| Input per 1M tokens | $2.10 | $0.1575 Lower |
| Cached input per 1M tokens | $0.2625 | $0.0168 |
| Cache write per 1M tokens | Same as input | Same as input |
| Output per 1M tokens | $6.30 | $0.4935 Lower |
| Context window | 262,144 tokens | 1,000,000 tokens Larger |
| Max output | 128,000 tokens | 131,072 tokens Larger |
| Capabilities | Reasoning, tool use, vision | Reasoning, tool use, vision |
| Takes | Text, image | Text, image |
| Returns | Text | Text |
| Zero data retention | Yes, on every provider | Yes, on every provider |
| No training on requests | Yes, on every provider | Yes, on every provider |
| Released | 2026-08-02 | 2026-09-17 |
Example costs
Example requests, priced at today’s rates with no cached input.
| Example request | Qwen 3.8 Max | Qwen 3.8 Omni Flash |
|---|---|---|
| 1,000 tokens in, 200 out | $0.00336 | $0.000256 |
| 10,000 tokens in, 1,000 out | $0.0273 | $0.002069 |
| 100,000 tokens in, 4,000 out | $0.2352 | $0.017724 |