News - Cocoloop
HomeClaudeOpenAIGeminiDeepSeekOpen SourceAll TagsArchive
文/A EN ▾
简体中文 Simplified Chinese English English 日本語 Japanese 한국어 Korean 繁體中文 Traditional Chinese Bahasa Indonesia Indonesian Tiếng Việt Vietnamese Deutsch German Português Portuguese Español Spanish Français French

Page 24

View archive →
Google 2026-09-20

UN Teams Up With Google to Rebuild Data Portal, Adds MCP Support

UNICEF tested six AI models on development data and got 21.2% average accuracy. The UN and Google just added MCP access to fix it.

#Open data#MCP
Qwen 2026-09-20

Alibaba's New Live-Translation Model Cuts Delay to 2.3 Seconds

Qwen3.8-LiveTranslate trims per-character latency from 2.8 to 2.3 seconds, understands 60 languages and speaks 29, with a demo page and API now live.

#Alibaba#voice AI#multimodal AI
Anthropic 2026-09-20

Anthropic Said to Prep New Model Before IPO to Counter Astra

Reuters says Anthropic may release a new model ahead of its expected IPO to respond to OpenAI's GPT-6 Astra, though no name or timeline has been confirmed.

#IPO#LLM competition
StepFun 2026-09-20

StepFun Launches Step 5 Preview, Claims Cost 1/8 of Opus 5

StepFun's new Step 5 Preview MoE model claims a top-three open-weight benchmark score and a cost one-eighth of Claude Opus 5, with full weights due October 15.

#Open Source#Large Models#MoE architecture
OpenAI 2026-09-20

OpenAI Launches Astra for Law, Scores 54% on Legal Research

OpenAI's Astra for Law pairs GPT-6 Astra with a self-built US legal search index, hitting 54.0% accuracy on a 200-question benchmark versus 38.7% for web-search-only Astra.

#Legal AI#Enterprise AI
Military AI 2026-09-20

US military nearly boarded Chinese ship over AI-fabricated intel

A US Pacific Command analyst used a chatbot to check a Chinese ship's cargo, then AI drafted the intel report that nearly triggered a boarding raid.

#AI Safety#AI Incidents#United States
Anthropic 2026-09-18

Claude Speeds Up 30 Biomolecular Models by About 4x

Anthropic had Claude optimize inference for 30-plus open-source biomolecular models, cutting runtime by about 4x on average with minimal accuracy loss; code is on GitHub.

#AI drug discovery#Claude#Open Source
GitHub Copilot 2026-09-18

GitHub Rewrote Its Copilot Runtime in 832,000 Lines of Rust

GitHub says Copilot's agent runtime went from roughly 430,000 lines of TypeScript to 832,378 lines of Rust in 14.5 weeks, led mostly by one engineer and AI agents.

#Rust#AI Coding
AI copyright 2026-09-18

Microsoft Exec Called AI Training the Biggest Labor Theft Ever

An unsealed NYT legal brief quotes a Microsoft director calling AI training on news content the ‘largest theft of labor in history.’

#OpenAI#Microsoft
US-China AI 2026-09-18

Federal Register Quietly Pulls Alibaba's Qwen-Powered Search

The US Federal Register quietly pulled an Alibaba Qwen-powered search feature this week, right as an FBI claim and a US-China summit both loom.

#Qwen#AI Regulation
← PreviousPage 24 of 159Next →

News · Cocoloop

AI news and analysis covering frontier models, open-source communities and industry moves, edited by Cocoloop from verified public sources.

Model News

  • Claude
  • OpenAI
  • Gemini
  • DeepSeek
  • Qwen

Topics

  • Open Source
  • AI Coding
  • Agent
  • All Tags

Site

  • Home
  • Archive
  • RSS Feed
  • robots.txt
  • Editorial Standards

Links

  • Cocoloop Main Site
  • Q&A Site
  • Hermes Guide
  • PixPix AI Images
  • AI Image Community
文/A EN ▾
简体中文 Simplified Chinese English English 日本語 Japanese 한국어 Korean 繁體中文 Traditional Chinese Bahasa Indonesia Indonesian Tiếng Việt Vietnamese Deutsch German Português Portuguese Español Spanish Français French

© 2026 News · Cocoloop — Frontier AI news

Sitemap