Qwen 3.6 27B vs Qwen 3.8 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 3.6 27Balibaba/qwen3.6-27b | Qwen 3.8 Flashalibaba/qwen3.8-flash |
|---|---|---|
| Maker | alibaba | alibaba |
| Kind | Chat | Chat |
| Input per 1M tokens | $0.63 | $0.1575 Lower |
| Cached input per 1M tokens | Not applicable | $0.0168 |
| Cache write per 1M tokens | Not applicable | $0.21 |
| Output per 1M tokens | $3.78 | $0.4935 Lower |
| Context window | 256,000 tokens | 991,000 tokens Larger |
| Max output | 256,000 tokens Larger | 128,000 tokens |
| Capabilities | Reasoning, tool use, vision, file input | Reasoning, tool use, vision, file input |
| Takes | Text, image, pdf | Text, image, pdf |
| 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 | 2026-04-22 | 2026-08-26 |
Example costs
Example requests, priced at today’s rates with no cached input.
| Example request | Qwen 3.6 27B | Qwen 3.8 Flash |
|---|---|---|
| 1,000 tokens in, 200 out | $0.001386 | $0.000256 |
| 10,000 tokens in, 1,000 out | $0.01008 | $0.002069 |
| 100,000 tokens in, 4,000 out | $0.07812 | $0.017724 |