Katteb

Hy4 Model to analyze million-token research documents

Tencent Chat and reasoning models Upgrade to use tencent/hy4-preview

Hy4 preview is Tencent Hunyuan’s new open-weight Mixture-of-Experts model with a context window exceeding one million tokens and strong Chinese and English ability. It is suited for long-form research threads, large codebases, and multi-document analysis, and is available in Katteb Chat at 6 credits per message.

Use Hy4 now All models

What it can do

  • Searches the live web and cites sources
  • Thinks step by step on hard questions
Takestext
Returnstext

What people use it for

Long-form research summarization

Feed a full research paper set or a month of chat logs and get a single structured summary with references and next-step suggestions. Hy4’s million-token context keeps citations and cross-references inside the same session so the summary can reference exact passages.

Large codebase inspection and explanation

Upload large repositories or multi-file projects and ask Hy4 to find bugs, explain modules, or produce a testing checklist. The model’s scale and co-design with Tencent developer tools make it practical for multi-file reasoning across long contexts.

Cross-language document analysis

Run bilingual analysis, extract structured data from Chinese and English reports, and produce aligned outputs in both languages. Hy4 is specifically developed with strong Chinese and English performance and is offered open-weight for integration.

Why it is worth it

Questions people ask

What is Hy4 preview good for?

Hy4 preview is built for very long-context reasoning and multi-document work: research papers, large codebases, and agentic productivity tasks. It shows particular strength in Chinese and English and is offered as an open-weight preview by Tencent.

How does Katteb charge for Hy4 messages?

Katteb bills Hy4 usage per message in credits. Hy4 messages cost 6 credits each; usage is measured on actual token consumption per message rather than a flat dollar price.

How do I keep long context across model switches?

Katteb Chat gives one account access to multiple models side by side and preserves the conversation context when you switch models mid-chat. That lets you keep the million-token thread while trying Hy4 and other models.

Related models