Urgent.News

What's breaking now, across thousands of outlets.

AI

Tencent unveils AI model it says outperforms Z.ai, Moonshot

Tencent demonstrated Hy4 Preview for game creation, 3D web design, and accounting compliance.

Tencent unveils AI model it says outperforms Z.ai, Moonshot

China's Tencent has unveiled a preview of a new open-source AI model named Hy4 preview, designed for tasks like software engineering, research and financial analysis, according to a Friday announcement on Hugging Face. This model employs a mixture-of-experts design with a total of 770 billion parameters, though only around 49 billion are utilized for each text request.

Tencent plans to integrate Hy4 preview with its existing products, including CodeBuddy and WorkBuddy. The company notes that the early-release model may occasionally take longer than expected to process intricate queries and may over-verify its own responses. This release comes as Tencent intensifies its AI investment in a competitive market, recently launching Hunyuan 3.0 in April, marking its first major model since hiring former OpenAI researcher Yao Shunyu to spearhead its AI platform development.

Written by urgent.news from The Indian Express's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

This story

This is one outlet's version. Read the fullest account.

Read the original at techinasia.com →

More in AI

RAG Evaluation: How to Know Your Retrieval Isn't Broken

The metric taxonomy for RAG — retrieval metrics, generation metrics, the eval sets you build, and the numbers that tell you which layer is failing.

  • Retrieval layer returning correct chunk only 30% of the time
  • Metric taxonomy includes retrieval and generation metrics
  • Hit rate measures fraction of queries with correct document in top-k results

'Enshittified at a Surprising Clip': A Week of Hacker News on AI Coding Tools

Some fortnights the complaints about AI come from people who barely use it. This one they came from the people who use it most.

  • Developers on Hacker News express frustration with AI coding assistants.
  • Cursor criticized for intrusive behavior and frequent UI changes.
  • GitHub Copilot billing issue and Codex bug on AWS Bedrock highlighted.

GitHub Copilot Spending Limit: How to Set It, What It Caps

A GitHub Copilot spending limit is a monthly budget, set in billing settings, that caps metered AI credit consumption for an enterprise, an organization, a cost center, or a single user.

  • GitHub Copilot spending limit is a monthly budget for enterprise, organization, cost center or user
  • Default setting disables usage blocking, making budget an alert rather than a limit
  • Budget caps metered AI credits, not seat cost, because seats are license-based

More from Saturday 29 August →