---
last_updated: 2026-11-21
source_url: https://opper.ai/tencent/hy4-preview
title: "Hy4 Preview by Tencent"
description: "Tencent Hy4 preview is Tencent Hy's open-weight 770B MoE model with 49B active parameters and a 1M-token context, built for long-horizon coding, agentic and office work."
---

# Hy4 Preview

By [Tencent](https://opper.ai/tencent).

Tencent Hy4 preview is a large language model from Tencent's Hy (Hunyuan) team, released on August 28, 2026 under the Apache 2.0 license. It is a Mixture-of-Experts model with 770 billion total parameters and 49 billion active per token, and it supports a context window of 1 million tokens.

The backbone has 78 layers: the first uses a dense feed-forward block and the other 77 use 256 routed experts (top-8) plus one shared expert. Attention is Gated DeepSeek Sparse Attention with IndexCache for reusing sparse indices across layers, the network uses four residual streams, and a native multi-token-prediction layer supports speculative decoding. Reasoning defaults to a high-effort chain of thought and can be switched to a no-think mode for direct answers, and the model supports tool calling.

On Tencent's numbers it scores 92.3 on GPQA Diamond, 65.7 on SWE-Bench Pro, 64.3 on Deep SWE, 74.1 on Toolathlon and 85.4 on Terminal-Bench 2.1. In a blind evaluation by 163 experts on 203 engineering tasks, Tencent reports an average of 2.99 out of 4 against 2.92 for GLM-5.3 and 2.94 for Kimi K3. It fits software engineering, office and analytical work, and scientific research tasks that need long context.

## Available routes

Hy4 Preview runs on 3 different routes through the Opper gateway. Prices are per 1M tokens. Its routes and prices, live from the catalogue, are public JSON with no API key needed at https://api.opper.ai/v3/models?q=hy4-preview.

| Provider | Model id | Region | Zero data retention | Training | Input | Output | Cache read |
| --- | --- | --- | --- | --- | --- | --- | --- |
| [DeepInfra](https://opper.ai/provider/deepinfra) | `deepinfra/tencent/Hy4-preview` | US (United States) | Yes | No | $0.83 | $2.50 | $0.04 |
| [Novita](https://opper.ai/provider/novita) | `novita/tencent/hy4-preview` | US (United States) | Yes | No | $0.83 | $2.50 | $0.04 |
| [Tencent Cloud](https://opper.ai/provider/tencent-cloud) | `tencent:sg/hy4-preview` | Multi-region | Yes | No | $0.83 | $2.50 | $0.04 |

## Key info

- Model id: `hy4-preview` (routes across Novita, Tencent Cloud)
- Input: Text
- Output: Text
- Features: Reasoning, Tools
- Context window: 1M
- Max output: 64K
- Input price: $0.83 per 1M tokens ($0.04 cache read)
- Output price: $2.50 per 1M tokens
- US residency available on DeepInfra and Novita
- Zero data retention on pay-as-you-go
- No training by default

## Data handling per route

Each route hosting Hy4 Preview has its own privacy posture, residency, and GDPR terms. Postures are maintained by Opper with a last-verification timestamp.

### DeepInfra — United States

Zero data retention: nothing is logged, held for abuse monitoring or used for training. No training on customer data. US; SCCs.

- Zero data retention: Yes. Nothing is logged, held for abuse monitoring or used for training.
- Training: No training on customer data.
- Logging: None
- Abuse monitoring: No classifier
- Transfer mechanism: SCCs

Verified by Opper on 2026-10-19.

### Novita — United States

Zero data retention: nothing is logged, held for abuse monitoring or used for training. No training on customer data. US; SCCs.

- Zero data retention: Yes. Nothing is logged, held for abuse monitoring or used for training.
- Training: No training on customer data.
- Logging: None
- Abuse monitoring: On by default, keeps nothing
- Transfer mechanism: SCCs

Verified by Opper on 2026-09-15.

### Tencent Cloud — Multi-region

Zero data retention: nothing is logged, held for abuse monitoring or used for training. No training on customer data. GLOBAL; SCCs.

- Zero data retention: Yes. Nothing is logged, held for abuse monitoring or used for training.
- Training: No training on customer data.
- Logging: None
- Abuse monitoring: On by default, keeps nothing
- Caching: Content cached for replay
- Transfer mechanism: SCCs

Verified by Opper on 2026-10-06.

## Get started

Call Hy4 Preview through the Opper gateway with one API key. Opper is drop-in compatible with the OpenAI, Anthropic and Google AI SDKs.

Hy4 Preview is a premium model on Opper. Sign up needs no credit card: you get an API key straight away and the free models work in the playground and the API. Add a card to use premium models, pay-as-you-go with no minimum.

### Set it up with your agent

Paste this into a coding agent like Claude Code, Cursor or Codex and it will wire up Opper for you:

```text
Use curl to download, read and follow: https://skills.opper.ai Call Hy4 Preview
```

### Or call it directly

OpenAI:

```javascript
import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.OPPER_API_KEY,
  baseURL: "https://api.opper.ai/v3/compat",
});

const completion = await client.chat.completions.create({
  model: "hy4-preview",
  messages: [{ role: "user", content: "Hello" }],
});
console.log(completion.choices[0].message.content);
```

Anthropic:

```javascript
import Anthropic from "@anthropic-ai/sdk";

const client = new Anthropic({
  apiKey: process.env.OPPER_API_KEY,
  baseURL: "https://api.opper.ai/v3/compat",
});

const message = await client.messages.create({
  model: "hy4-preview",
  max_tokens: 1024,
  messages: [{ role: "user", content: "Hello" }],
});
console.log(message.content);
```

Gemini:

```javascript
import { GoogleGenAI } from "@google/genai";

const ai = new GoogleGenAI({
  apiKey: process.env.OPPER_API_KEY,
  httpOptions: { baseUrl: "https://api.opper.ai/v3/compat" },
});

const response = await ai.models.generateContent({
  model: "hy4-preview",
  contents: "Hello",
});
console.log(response.text);
```

cURL:

```bash
curl https://api.opper.ai/v3/compat/chat/completions \
  -H "Authorization: Bearer $OPPER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "hy4-preview",
    "messages": [{ "role": "user", "content": "Hello" }]
  }'
```

## Use Hy4 Preview in your apps

Pick your tool to see how to run it on Hy4 Preview through Opper.

- [Claude Code](https://opper.ai/apps/claude-code/hy4-preview)
- [Claude Cowork](https://opper.ai/apps/claude-cowork/hy4-preview)
- [Codex](https://opper.ai/apps/codex/hy4-preview)
- [DeepSeek Harness](https://opper.ai/apps/deepseek-harness/hy4-preview)
- [GitHub Copilot CLI](https://opper.ai/apps/github-copilot-cli/hy4-preview)
- [Aider](https://opper.ai/apps/aider/hy4-preview)
- [Goose](https://opper.ai/apps/goose/hy4-preview)
- [Droid](https://opper.ai/apps/droid/hy4-preview)
- [OpenCode](https://opper.ai/apps/opencode/hy4-preview)
- [Pi](https://opper.ai/apps/pi/hy4-preview)
- [OpenClaw](https://opper.ai/apps/openclaw/hy4-preview)
- [Hermes](https://opper.ai/apps/hermes/hy4-preview)
- [Copilot for Obsidian](https://opper.ai/apps/obsidian-copilot/hy4-preview)
- [OmniRoute](https://opper.ai/apps/omniroute/hy4-preview)
- [Cursor](https://opper.ai/apps/cursor/hy4-preview)
- [Cline](https://opper.ai/apps/cline/hy4-preview)
- [Continue](https://opper.ai/apps/continue/hy4-preview)
- [Kilo Code](https://opper.ai/apps/kilo-code/hy4-preview)
- [GitHub Copilot](https://opper.ai/apps/github-copilot/hy4-preview)
- [Zed](https://opper.ai/apps/zed/hy4-preview)
- [JetBrains AI Assistant](https://opper.ai/apps/jetbrains-ai-assistant/hy4-preview)
- [Voidleap Code](https://opper.ai/apps/voidleap-code/hy4-preview)
- [n8n](https://opper.ai/apps/n8n/hy4-preview)
- [LibreChat](https://opper.ai/apps/librechat/hy4-preview)
- [Open WebUI](https://opper.ai/apps/open-webui/hy4-preview)
- [See all apps](https://opper.ai/apps)

## Other models from Tencent

- [Hy4](https://opper.ai/tencent/hy4)
- [Hy3](https://opper.ai/tencent/hy3)
- [Hy Image 3.5 Preview](https://opper.ai/tencent/hy-image-3-5-preview)
