The Qwen3.5 122B-A10B 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.
function_calling| Model | Context | Max out | Pricing |
|---|---|---|---|
| Qwen: Qwen3.5-Flash | 1M | 66K | $0.07$0.26 |
| Qwen: Qwen3.5-35B-A3B | 262K | 82K | $0.14$1 |
| Qwen: Qwen3.5-9B (batch) | 262K | 236K | $0.17$0.25 |
| Qwen: Qwen3.5-9B | 262K | 236K | $0.17$0.25 |
| Qwen: Qwen3.5-122B-A10B | 262K | 66K | $0.26$2.08 |
| Qwen: Qwen3.5-27B | 262K | 82K | $0.26$2.6 |
| Qwen: Qwen3.5 Plus 2026-02-15 | 1M | 66K | $0.26$1.56 |
| Qwen: Qwen3.5 Plus 2026-04-20 | 1M | 66K | $0.3$1.8 |
| Qwen: Qwen3.5 397B A17B | 262K | 236K | $0.5$3.6 |
/v1/models/openrouter/qwen/qwen3.5-122b-a10bThe Qwen3.5 122B-A10B 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.