Qwen vision-language thinking model for visual reasoning, documents, and agent tasks
Qwen3 VL 235B A22B Thinking by alibaba is offered by 11 providers on this page. Its reference price is $0.50 per 1M input tokens and $6.00 per 1M output tokens. The lowest paid channel is Merge Gateway at $0.287 / $2.87 per 1M, about 3.4× below the reference price. That channel is a third-party gateway, so confirm its availability and rate limits before depending on it.
The context window is 131,072 tokens, with an output limit of 32,768 tokens. It supports reasoning, tool use, and structured output. Accepted input modalities are Text, Image, and Video. The weights are open, so it can also be self-hosted or served through a gateway of your choice. Providers report the capability flags inconsistently, so verify a specific feature against the host you plan to use. Its training knowledge cuts off at 2025-03-31.
| Provider | Tier | Input | Output | Cache read | Cache write | Context | Output limit | Status |
|---|---|---|---|---|---|---|---|---|
| Merge Gateway | Gateway | $0.29 | $2.87 | $0.057 | — | 131,072 | 32,768 | |
| Kilo Gateway | Gateway | $0.40 | $4.00 | — | — | 131,072 | 32,768 | |
| OpenRouter | Gateway | $0.40 | $4.00 | — | — | 131,072 | 32,768 | |
| OrcaRouter | Gateway | $0.40 | $4.00 | — | — | 131,072 | 40,960 | |
| Eden AI | Gateway | $0.40 | $4.00 | — | — | 131,072 | 32,768 | |
| NovitaAI | Gateway | $0.98 | $3.95 | — | — | 131,072 | 32,768 | |
| Hugging Face Qwen/Qwen3-VL-235B-A22B-Thinking | Cloud | $0.98 | $3.95 | — | — | 131,072 | 32,768 | |
| Jalapeno Cloud Qwen3-VL-235B-A22B-Thinking | Gateway | $0.98 | $3.95 | — | — | 131,072 | 32,768 | |
| DevPass (LLM Gateway) | Gateway | $0.98 | $3.95 | — | — | 131,072 | 8,192 | |
| LLM Gateway | Gateway | $0.98 | $3.95 | — | — | 131,072 | 32,768 | |
| NanoGPT | Gateway | $0.50 | $6.00 | $0.25 | — | 131,072 | 32,768 |
Sorted by blended price (input×0.75 + output×0.25) asc. The official channel always shows regardless of rank. Whether a gateway's low price is actually usable can't be verified.
Interleaved thinking (reasoning between tool calls) is declared by 1 of 11 providers.
7 / 11 providers expose no reasoning control (reasoning_options: []).