:

TENCENT USED ANTHROPIC'S CLAUDE TO BUILD HY3 MODEL

AI DESK2 MIN READ
TUE, APR 28, 2026

■ AI-SUMMARIZED FROM 1 SOURCE BELOW

Tencent employees leveraged Anthropic's Claude Code to evaluate and fine-tune the Chinese tech giant's new Hy3 AI model, according to internal sources and memos reviewed by The Information.

Tencent's latest AI model has received positive feedback from developers, but internal documents suggest the company relied on Anthropic's Claude Code tool during the development process. According to sources and memos obtained by The Information, Tencent staff used Claude Code to assist with evaluating the Hy3 model's performance and refining its capabilities. The tool helped the company identify areas for improvement and optimize the model's outputs. The disclosure reveals Tencent's reliance on third-party AI development tools from Anthropic, a competitor in the generative AI space. While Tencent has invested heavily in building its own AI capabilities, the use of Claude Code suggests the company found value in external resources for model optimization. Hy3 has garnered developer interest since its release, with positive reviews highlighting its performance improvements. The model represents Tencent's push to compete in the rapidly evolving AI market dominated by players like OpenAI, Google, and others. The story underscores the interconnected nature of the AI industry, where even major tech companies leverage tools and platforms from competitors to advance their own projects. Anthropic, founded by former OpenAI researchers, has positioned Claude as a versatile AI assistant capable of handling coding tasks and technical evaluations. Tencent has not publicly commented on its use of Claude Code for Hy3 development. The company continues to invest in AI research and development as part of its broader strategy to strengthen its position in artificial intelligence.

■ SOURCES

Techmeme

■ SUMMARY WRITTEN BY AI FROM THE LINKS ABOVE

■ MORE FROM THE AI DESK

A large-scale analysis of archived websites reveals AI-generated content has already saturated the internet, creating unexpected side effects: homogenized writing and an unusual cheerfulness across the web.

1H AGOAI Desk

OpenAI rejected reports of missed internal targets on Tuesday, asserting that both its consumer and enterprise divisions are performing strongly. The statement comes as the AI startup faces scrutiny over its expansion pace.

1H AGOAI Desk

As AI agents gain the ability to make purchases on users' behalf, the FIDO Alliance has partnered with Google and Mastercard to develop safeguards preventing unauthorized or errant transactions.

1H AGOAI Desk

At least two China-based venture funds backing major AI companies have adopted parallel fundraising structures to attract US investors while circumventing American compliance restrictions.

1H AGOAI Desk

■ SUBSCRIBE TO THE DAILY BRIEF

ONE EMAIL, 5 STORIES, 06:00 UTC. UNSUBSCRIBE ANYTIME.