Qwen3.5 397B A17B
AlibabaThe Qwen3.5 series 397B-A17B native vision-language model is built on a hybrid architecture that integrates a linear attention mechanism with a sparse mixture-of-experts model, achieving higher inference efficiency. It delivers...
| Route | Entrée | Sortie | Blended | vs |
|---|---|---|---|---|
| ● Alibaba le moins cher | $0.39 | $2.34 | $0.88 | - |
| $0.45 | $3.00 | $1.09 | +24% | |
| $0.50 | $3.60 | $1.27 | +44% | |
| $0.55 | $3.50 | $1.29 | +47% | |
| ● Phala | $0.55 | $3.50 | $1.29 | +47% |
| ● AtlasCloud fp8 | $0.55 | $3.50 | $1.29 | +47% |
| ● StreamLake | $0.60 | $3.60 | $1.35 | +53% |
| $0.60 | $3.60 | $1.35 | +53% | |
| $0.60 | $3.60 | $1.35 | +53% | |
| ● Venice | $0.75 | $4.50 | $1.69 | +92% |
● API officielle · ● plateforme cloud · ● hébergeur d'inférence · USD par million de tokens. Même modèle, fournisseurs différents - le prix et la quantization varient.
- Ctx
- 262K
- max output
- 236K
- cache read /1M
- $0.30
- Famille
- qwen
- Modalités
- text · image · video
- Fonctions
- json · reasoning · tools
Prix live, mis à jour le 2026-11-03.
| Modèle | Entrée | Sortie | Ctx |
|---|---|---|---|
| Qwen3.6 Plus | $0.325 | $1.95 | 1M |
| Qwen3.5-27B | $0.26 | $2.6 | 262K |
| Qwen3.6 27B | $0.45 | $2.7 | 262K |
| Qwen3 235B A22B Thinking 2507 | $0.45 | $3.5 | 128K |
| Qwen3.5 397B A17B · ce modèle | $0.5 | $3.6 | 262K |
| Qwen3.7 Max | $1.48 | $4.42 | 1M |