# Qwen3.5-397B-A17B

> Open-weight native vision-language model from Alibaba, built on linear attention and sparse experts.

## Key facts

- Model author: Qwen
- Input: $0.60 per 1M tokens
- Output: $3.60 per 1M tokens
- Context window: 256,000 tokens
- Maximum output: 8,000 tokens
- Input modalities: image, text, video
- Output modalities: text
- Capabilities: reasoning, structured_output, tools, vision
- Released: 2026-02-16
- Model IDs: `qwen/qwen3.5-397b-a17b`
- Model page: https://www.realrelay.ai/model/qwen/qwen3.5-397b-a17b

## Overview

Qwen3.5-397B-A17B is a native vision-language model from the Alibaba Qwen3.5 series with open weights. Its hybrid architecture integrates a linear attention mechanism with a sparse mixture-of-experts model for higher inference efficiency, and it takes image, text and video input and returns text.

It has a 256K-token context window and can generate up to 8,000 output tokens. It accepts text, image and video input and returns text. It was released on February 16, 2026. Its weights are open.

## Capabilities

Qwen3.5-397B-A17B supports tool calling, reasoning, structured output and vision. It accepts text, image and video input and returns text.

Being a native vision-language model, it handles images, video and text in one model. Alibaba documents function calling for the series, so it can act on what it sees, and structured output in non-thinking mode for valid JSON from visual inputs.

## Pricing

Qwen3.5-397B-A17B costs $0.60 per 1M input tokens and $3.60 per 1M output tokens. Qwen3.5-397B-A17B is available through RealRelay's OpenAI-compatible API.

## Use cases

Use Qwen3.5-397B-A17B for multimodal agents that look at screenshots, photos or video, reason about them and call tools in response.

Because the weights are open, it also fits teams that want to prototype against the hosted API and keep the option of running the model in their own infrastructure, with structured output for parseable results.

## Call it

OpenAI-compatible chat completions:

```bash
curl https://www.realrelay.ai/v1/chat/completions \
  -H "Authorization: Bearer $REALRELAY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "qwen/qwen3.5-397b-a17b", "messages": [{"role": "user", "content": "Hello"}]}'
```

## FAQ

### What is Qwen3.5-397B-A17B?

Qwen3.5-397B-A17B is a native vision-language model from the Alibaba Qwen3.5 series with open weights.

### How much does Qwen3.5-397B-A17B cost?

Qwen3.5-397B-A17B costs $0.60 per 1M input tokens and $3.60 per 1M output tokens.

### What is the context window of Qwen3.5-397B-A17B?

Qwen3.5-397B-A17B has a 256K-token context window and can generate up to 8,000 output tokens.

### What inputs and outputs does Qwen3.5-397B-A17B support?

Qwen3.5-397B-A17B accepts text, image and video input and returns text.

### Does Qwen3.5-397B-A17B support tool calling?

Yes. Qwen3.5-397B-A17B supports tool calling, reasoning, structured output and vision.

### When was Qwen3.5-397B-A17B released?

Qwen3.5-397B-A17B was released on February 16, 2026.

### How do I call Qwen3.5-397B-A17B via the API?

Send requests to RealRelay's OpenAI-compatible API with the model ID qwen/qwen3.5-397b-a17b to use Qwen3.5-397B-A17B.