Updated daily488 verified stories
Models News tracker.
New frontier, open-weight, multimodal, reasoning, and specialist AI models—plus the evidence behind their capability claims.
- 38stories this week
- 488sources cited
- 488verified timeline entries
Verified sourcing
Every story links to the strongest available evidence: original sources when available, otherwise clearly attributed reporting.
Plain English
What happened, why it matters, and what to watch — without the jargon.
No filler
When the signal is thin, we publish nothing rather than padding the feed.
More stories
9 storiesEnterprise
Meta and Microsoft slash Claude spending, pivot to in‑house AI tools
Meta and Microsoft, two of Anthropic’s biggest enterprise customers, have sharply reduced internal use of Claude, cutting budgets by more than a third and shifting to their own AI offerings such as GitHub Copilot, Muse Code and MetaCode.the-decoder.comPolicy
Pentagon confirms end to Anthropic AI use after supply chain risk designation
The US Department of Defence has officially ceased using Anthropic's AI tools, confirming a phase-out ordered in February after the company refused to remove safety guardrails from its Claude models.bbc.comProduct
Qwen Code Desktop 0.25.0 adds Linux aarch64 support and new desktop controls
The latest Qwen Code Desktop release (v0.25.0) expands platform coverage to Linux aarch64, introduces an opt‑out flag for automatic updates, and adds several usability and performance tweaks for the AI‑powered coding environment.github.comPolicy
South Korea's science ministry reverses course on AI project support
The Ministry of Science and ICT retracted a plan to overhaul its national AI foundation model project just hours after suggesting it would end support for selected teams.biz.chosun.comIndustry
US lead in AI over China narrows to record low after DeepSeek gains
Bloomberg Intelligence reports that the performance gap between top US and Chinese AI models has shrunk to 3%, driven by the release of DeepSeek’s V4.1 Flash.bloomberg.comInnovation
Navdyut AI Labs open-sources 240M parameter model for AMD hardware
Navdyut AI Labs, a bootstrapped lab in Assam, has released a 240-million-parameter generative foundation model trained from scratch, marking the first such release from Northeast India and highlighting a strategy to reduce reliance on Nvidia's CUDA ecosystem.english.loktej.comProduct
Fulcrum releases Echo, an AI model designed to mimic specific writing styles
Startup Fulcrum has launched Echo, a large language model capable of generating text in the style of specific authors, raising questions about the ethics of automated stylistic imitation.theregister.comIndustry
Google releases Gemini 4 Argon model for complex enterprise and security workflows
Google has launched Gemini 4 Argon, its most advanced AI model to date, targeting complex software engineering, cybersecurity, and enterprise knowledge tasks.benzinga.comProduct
DeepSeek Harness adds experimental Claude Code Mods compatibility layer
DeepSeek released Harness v0.2.1-alpha.1, introducing an experimental compatibility layer that allows Claude Code Mods to run within its plugin system, demonstrating cross-tool extension reuse.eu.36kr.com
One useful briefing each week
Keep up with AI without living in the feed.
Get the week’s verified AI news, original data, useful tools, learning picks, and fresh AI jobs.
Reach people who are learning AI
Hiring an AI professional or launching a useful AI product? Put it in front of people who came here to learn and act.
Post an AI jobSubmit an AI tool