Qwen3.5-122B-A10B
Qwen released 2026-02-25
MiniMax M3 is both better and cheaper.
It scores +12.6 higher and costs 27% less ($0.525/M against $0.715/M) on this workload — and it does everything this model does.
DeepSeek V4 Flash 0731 is cheaper still (85% less) but drops no image, video input.
Every price dimension
| Input | $0.260/M |
|---|---|
| Output | $2.08/M |
Independent scores
| Intelligence | 32.8 |
|---|---|
| Coding | 45.7 |
| Agentic | 21.3 |
| LMArena Elo | 1417.9default |
Sources are listed separately rather than averaged. Across the 62 models both have scored they correlate at r = 0.806 — close agreement overall, but four models rank very differently between them, and a blended score would hide exactly those.
Capability
| Context window | 262K |
|---|---|
| Max output | 66K |
| Input modes | text, image, video |
| Tool use | yes |
| Reasoning | optional |
Provenance
| Price source | openrouter.ai |
|---|---|
| Fetched | 2026-08-24 |
| Quality data | verified |
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. In terms of...