Overview
Tencent Hy is Tencent's global foundation-model brand, formerly known as Tencent Hunyuan. The lineup spans language, reasoning, agent, image, video, and 3D models. Its current flagship language release is Hy4 Preview, released and open-sourced on August 28, 2026.
Hy4 Preview is a Mixture-of-Experts language model with 770 billion total parameters, 49 billion active parameters per token, and a 1M-token context window. Tencent positions it for real productivity tasks across coding, office work, game development, finance, security, and scientific research. It is available through Tencent products, Tencent Cloud TokenHub, OpenRouter, and self-hosted model weights.
For teams evaluating an AI agent or coding model, the important decision is not only benchmark score. Hy4 Preview combines Apache 2.0 open weights, a much longer context window than Hy3, hosted API access, and Tencent product integration. It is most relevant to developers who want to test a large open frontier model while keeping Hy3 as the safer stable baseline.
Key Features
770B MoE architecture - Hy4 Preview uses a mixture-of-experts design with 770B total parameters and 49B active parameters. Sparse activation is intended to improve serving efficiency compared with activating an equivalent dense model for every token.
1M-token context window - The model supports long documents, codebases, agent traces, research corpora, and multi-file reasoning tasks that exceed ordinary chat limits. This is useful for coding agents, long-form analysis, and enterprise knowledge work.
Productivity-focused training - Tencent says Hy4 Preview was developed with training data co-created by internal experts across software engineering, game development, finance, security, and related domains.
Tencent ecosystem integration - Hy4 Preview is available through WorkBuddy, CodeBuddy, Yuanbao, ima, and other Tencent products. This gives enterprise users more routes to evaluate the model beyond a standalone model card.
TokenHub and OpenRouter API access - Hy4 Preview can be reached through Tencent Cloud TokenHub and OpenRouter. Tencent's launch announcement lists API pricing at USD 0.834 per million input tokens, USD 2.501 per million output tokens, and USD 0.042 per million cache-hit tokens.
Apache 2.0 open weights - Hy4 Preview and Hy4 Preview-FP8 weights are published through Tencent's official model channels under Apache License 2.0, with vLLM and SGLang deployment recipes.
How It Compares
Hy4 Preview competes with Chinese and international frontier models used for reasoning, coding, and agent workflows. Its strongest differentiators are the combination of large MoE scale, 1M context, Apache 2.0 weights, Tencent product access, and hosted API routes.
| Tencent Hy4 Preview | Tencent Hy3 | GPT-5.6 Sol | Claude Opus 4.8 | GLM-5 | |
|---|---|---|---|---|---|
| Primary positioning | Open flagship productivity model | Stable Tencent Hy release | Flagship OpenAI model | Closed frontier assistant model | Chinese frontier model |
| Parameters | 770B total, 49B active | 295B total, 21B active | Undisclosed | Undisclosed | Undisclosed |
| Context | Up to 1M | Up to 256K | Up to 1M in published evals | Up to 1M in some surfaces | Up to 1M in some surfaces |
| Hosted API | Tencent Cloud TokenHub, OpenRouter | Tencent Cloud TokenHub | OpenAI API | Anthropic API | Varies by provider |
| Open weights | Yes, Apache 2.0 | Yes, custom Tencent license | No | No | Partial/provider-dependent |
The main tradeoff is release maturity. OpenAI and Anthropic have broader global developer tooling and third-party integrations, while Hy3 has a more stable Tencent release profile. Tencent Hy4 Preview may be more attractive when cost, 1M context, Apache 2.0 weights, Chinese-language ecosystem alignment, or Tencent Cloud deployment matter more.
Pricing & Plans
Tencent Hy has multiple access routes: product integrations, TokenHub API access, OpenRouter access, and model materials for self-hosted evaluation. API pricing is usage-based rather than a flat subscription.
Tencent's Hy4 Preview launch announcement lists:
| Token type | Official listed price |
|---|---|
| Input tokens | USD 0.834 per million tokens |
| Output tokens | USD 2.501 per million tokens |
| Cache hits | USD 0.042 per million tokens |
At launch, Hy4 Preview is free on WorkBuddy and CodeBuddy for two weeks, and free Hy3 access on both platforms has been extended until September 30, 2026. If you are buying through Tencent Cloud, OpenRouter, or another provider, verify the current region, currency, quota, cache policy, and billing unit before estimating production cost.
Best For
- Tencent Cloud or OpenRouter users who want a hosted route to Hy4 Preview.
- Developers building coding agents, office agents, search agents, or long-document workflows.
- Research and ML teams evaluating Apache 2.0 open frontier-class MoE models.
- Enterprises in supported Tencent regions that need Chinese ecosystem alignment.
- Cost-sensitive teams comparing hosted reasoning models across providers.
FAQ
What is Tencent Hy?
Tencent Hy is Tencent's global foundation-model brand, formerly known as Tencent Hunyuan. It covers language, reasoning, agent, image, video, and 3D models.
What is Hy4 Preview?
Hy4 Preview is Tencent's August 2026 language-model release built on a 770B-parameter MoE architecture with 49B active parameters and up to 1M context.
Is Hy4 Preview different from Hy3?
Yes. Hy4 Preview is larger than Hy3, raises context from 256K to 1M tokens, and uses Apache 2.0 weights. Hy3 remains the more stable official July 2026 release baseline.
How much does the Tencent Hy4 Preview API cost?
Tencent's launch announcement lists Hy4 Preview at USD 0.834 per million input tokens, USD 2.501 per million output tokens, and USD 0.042 per million cache-hit tokens.
Is Tencent Hy4 Preview open source?
Hy4 Preview is released under Apache License 2.0, with model weights available through official Tencent channels including Hugging Face, ModelScope, GitCode, and CNB.
What context length does Hy4 Preview support?
Tencent states Hy4 Preview supports up to 1M tokens of context. That is a major increase over Hy3's 256K-token context window.
Who should consider Tencent Hy?
Tencent Hy is most relevant for teams that need long-context reasoning, coding-agent capability, Tencent Cloud access, Chinese ecosystem alignment, or an open-weight alternative to closed API-only models.