Qwen3.5-122B-A10B
Qwen · Open weights · Released Feb 25, 2026 · qwen/qwen3.5-122b-a10b
262K contextVisionToolsReasoningStructured outputPrompt cachingOpen weights
Available from 5 providersUSD / MTOK · VERIFIED 4 MINUTES AGO
EST. MONTHLY COSTclick a row for detail · click headers to sort
| PROVIDER | INPUT | OUTPUT | CACHE | CONTEXT | QUANT | TTFT | TPS | SMOKE | UPTIME |
|---|---|---|---|---|---|---|---|---|---|
 SiliconFlow | $0.26 | $2.08 | — | 262K | fp8 | 1.9s | 39 tok/s | 2/5 | 97.7% |
 Alibaba | $0.26 | $2.08 | — | 262K | — | 1.5s | 160 tok/s | 5/5 | 94.1% |
 DeepInfra | $0.29 | $2.40 | — | 262K | fp4 | 0.75s | 123 tok/s | 3/5 | 99.9% |
| ATAtlasCloud | $0.30 | $2.40 | $0.30 | 262K | fp8 | 2s | 136 tok/s | 5/5 | 98.3% |
 Novita | $0.40 | $3.20 | — | 262K | bf16 | 1.7s | 151 tok/s | 4/5 | 99.7% |
TTFT/TPS: median (P50) over trailing 72h · chat10k profile (~10K in / 200 out) · measured via OpenRouter routing · SMOKE: of 5 programmatic capability probes passed (JSON schema, instruction following, tool call, long-context, benign compliance; retry-once) · screened via OpenRouter routing · v0.2 · methodology v1.1 · measured 43 hours ago
About
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...
Specifications
Context window262,144
Max output262,144
Modalities intext, image, video
Modalities outtext
LicenseOpen weights
ReleasedFeb 25, 2026
TokenizerQwen3
WeightsHugging Face ↗
Related
Embed badge
[](https://modelindex.ai/models/qwen/qwen3.5-122b-a10b)Sources
OpenRouter API4 minutes ago
AGGREGATORHugging Face4 minutes ago
COMMUNITYModelIndex benchmarks43 hours ago
MEASUREDVerified 4 minutes ago · list prices, not negotiated rates.