# Hy4 flash

> Hy4 flash is Tencent's model for everyday productivity, document understanding and agent workflows. It accepts text and images and produces text, with a 1M-token context window. Tencent recommends Hy4 for more complex, long-horizon software engineering.

**ID**: `tencent/hy4-flash`  
**Creator**: tencent  
**Category**: text  
**Context**: 1M tokens  
**Pricing**: $0.08 / $0.32 per 1M · billed by your provider (BYOK)  
**Released**: 2026-11-16  
**Web page**: https://anyrouter.dev/model/tencent/hy4-flash

**Input modalities**: text, image  
**Output modalities**: text  

**Capabilities**: chat, function-calling, vision, reasoning

## Usage

**Endpoint**: `POST https://anyrouter.dev/api/v1/chat/completions`

```bash
curl https://anyrouter.dev/api/v1/chat/completions \
  -H "Authorization: Bearer $ANYROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "tencent/hy4-flash",
  "messages": [
    {
      "role": "user",
      "content": "Say hi in 3 words."
    }
  ]
}'
```

## Providers

| Provider | Input | Output |
| --- | --- | --- |
| aihubmix-byok | $0.08/M (list) | $0.32/M (list) |

BYOK routes are billed by your own provider key — AnyRouter charges $0. Rates marked (list) are the provider's published list price.

## Model card

- Architecture: mixture of experts [1]
- Active parameters: 12B [1]
- Native context: 1048576 tokens [2]

### Strengths [1,2]

- Designed for high-frequency productivity and agent workflows.
- Native text and image understanding with a long context window.

### Known limits [1]

- Tencent recommends Hy4 for complex, long-horizon software engineering.
- Open weights were planned, not released, at launch.

### Use cases [1,2]

- Document and chart understanding
- Research and slide preparation
- Coding assistance and routine agent workflows

### API notes [1,2]

- Supports function calling, structured outputs, reasoning and prompt caching.
- Tencent documents Chat Completions, Responses and Messages protocols.
- Tencent reports 224B total parameters including 64B Engram parameters, with 12B active per token.

### Links

- announcement: https://hy.tencent.ai/research/hy4-flash
- docs: https://cloud.tencent.com/tokenhub/hy4-flash-en
- pricing: https://cloud.tencent.com/tokenhub/hy4-flash-en

### Sources

1. [Introducing Hy4 flash](https://hy.tencent.ai/research/hy4-flash) (accessed 2026-11-16)
2. [Hy4 flash model details](https://cloud.tencent.com/tokenhub/hy4-flash-en) (accessed 2026-11-16)

## Source

- Upstream docs: https://hy.tencent.ai/research/hy4-flash
