Qwen vision-language model for visual reasoning, documents, and agent tasks
Qwen/Qwen3-VL-8B-Instruct by alibaba is offered by 5 providers on this page. Its reference price is $0.18 per 1M input tokens and $0.68 per 1M output tokens. The lowest paid channel is NovitaAI at $0.08 / $0.50 per 1M, about 2.3× below the reference price. That channel is a third-party gateway, so confirm its availability and rate limits before depending on it.
Artificial Analysis rates it 7.3 on the Intelligence Index. Against its lowest blended price of $0.185 per 1M, that is roughly 39 index points per dollar, which is the value ratio the leaderboards rank on. Median output speed is 119 tokens per second, with 2.3s to the first token.
The context window is 262,000 tokens at the reference host, but hosts report different limits, from 131,072 to 262,144, so the usable window depends on the provider you pick. It supports tool use and structured output. Accepted input modalities are Text, Image, and Video. Providers report the capability flags inconsistently, so verify a specific feature against the host you plan to use.
Quality is independently evaluated by Artificial Analysis. Speed/latency are model-level medians.
| Provider | Tier | Input | Output | Cache read | Cache write | Context | Output limit | Status |
|---|---|---|---|---|---|---|---|---|
| NovitaAI | Gateway | $0.08 | $0.50 | — | — | 131,072 ⚠ | 32,768 | |
| OpenRouter | Gateway | $0.12 | $0.46 | — | — | 262,144 ⚠ | 32,768 | |
| Kilo Gateway | Gateway | $0.12 | $0.46 | — | — | 131,072 ⚠ | 32,768 | |
| SiliconFlow Qwen/Qwen3-VL-8B-Instruct | Cloud | $0.18 | $0.68 | — | — | 262,000 | 262,000 | |
| SiliconFlow (China) Qwen/Qwen3-VL-8B-Instruct | Cloud | $0.18 | $0.68 | — | — | 262,000 | 262,000 |
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.
Individual evaluations run by Artificial Analysis, on the same reasoning tier as the intelligence score above. Each benchmark has its own task set and harness, so rows are not comparable with one another. The Intelligence Index above draws on a different, newer set of evaluations.