Qwen3.5-35B-A3B pricing & benchmarks
Released Feb 25, 2026 · served by 9 providers
What Qwen3.5-35B-A3B is
The Qwen3.5 Series 35B-A3B is a native vision-language model designed with a hybrid architecture that integrates linear attention mechanisms and a sparse mixture-of-experts model, achieving higher inference efficiency. Its overall...
- Model ID:
qwen/qwen3.5-35b-a3b - Modality: text+image+video->text
- Tool calling: supported · Structured output: supported
- Weights:
Qwen/Qwen3.5-35B-A3Bon Hugging Face
Providers and prices for Qwen3.5-35B-A3B
The same model costs different amounts depending on who serves it. Prices are per 1M tokens, cheapest first. Uptime and throughput are OpenRouter's rolling measurements, not vendor claims.
| Provider | Input | Output | Context | Quant | Uptime 24h | Throughput |
|---|---|---|---|---|---|---|
| DeepInfracheapest | $0.140 | $1.00 | 262K | fp8 | 95.9% | — |
| AkashML | $0.140 | $1.00 | 262K | fp8 | 99.9% | — |
| Parasail | $0.150 | $1.00 | 262K | fp8 | 99.9% | — |
| CoreWeave | $0.250 | $1.25 | 262K | fp8 | 98.9% | — |
| Venice | $0.313 | $1.25 | 256K | — | 99.9% | — |
| Alibaba | $0.163 | $1.30 | 262K | — | 99.9% | — |
| NextBit | $0.230 | $1.60 | 262K | fp8 | 44.7% | — |
| AtlasCloud | $0.225 | $1.80 | 262K | fp8 | 100.0% | — |
| SiliconFlow | $0.240 | $1.80 | 262K | fp8 | 97.3% | — |
Cheaper models in the same class
Models scoring within 4 points of Qwen3.5-35B-A3B on the Intelligence Index, but with a lower output price.
FAQ
How much does the Qwen3.5-35B-A3B API cost?
$0.140 per 1M input tokens and $1.00 per 1M output tokens. A typical workload of 1M input + 300K output tokens costs about $0.44.
Is Qwen3.5-35B-A3B free?
No. It is a paid model starting at $0.140 per 1M input tokens, though some providers offer trial credits.
Which provider is cheapest for Qwen3.5-35B-A3B?
DeepInfra at $0.140 input / $1.00 output per 1M tokens (fp8 quantization).
Can I self-host Qwen3.5-35B-A3B?
Yes — weights are published as Qwen/Qwen3.5-35B-A3B on Hugging Face, so you can run it on your own hardware.