Anthropic lands $5B Amazon investment, vows $100B AWS spend
Amazon is investing another $5 billion in Anthropic as part of a circular AI partnership. In return, Anthropic has agreed to spend $100 billion on AWS cloud services.
TechCrunch ↗
NSA reportedly uses Anthropic’s restricted Mythos model
Reports say the NSA is using Anthropic’s restricted Mythos AI model even as tensions continue with the Pentagon. The story highlights government interest in advanced AI tools despite internal defense-sector friction.
TechCrunch ↗
Fermi CEO, CFO Exit Amid Texas AI Campus Headwinds
Fermi’s CEO and CFO have suddenly left the AI nuclear power startup. The company, co-founded by former U.S. Energy Secretary Rick Perry, has been facing challenges with its Texas AI campus plans.
TechCrunch ↗
Moonshot AI Unveils Kimi K2.6 Agentic Model
Moonshot AI released Kimi K2.6, an open-source multimodal agentic model aimed at long-running coding and software engineering tasks. The model is described as scaling to 300 sub-agents and 4,000 coordinated steps for autonomous workflows.
MarkTechPost ↗
Google expands Gemini in Chrome to seven countries
Google is rolling out Gemini in Chrome in seven new countries, including Australia, Indonesia, Japan, the Philippines, Singapore, South Korea, and Vietnam. The feature is available on desktop and iOS in all of these markets except Japan.
TechCrunch ↗
AI writing pattern becomes a clear tell
The article says a specific sentence pattern has become so common in AI-generated text that it is now almost a reliable sign the writing is synthetic. It highlights how this phrasing has spread enough to be recognized as a strong clue of AI authorship.
TechCrunch ↗
Qwen 3.6 tutorial covers multimodal AI workflows
This tutorial shows how to build an end-to-end implementation around Qwen 3.6-35B-A3B. It covers multimodal inference, thinking control, tool calling, MoE routing, RAG, and session persistence in a practical coding workflow.
MarkTechPost ↗
Phi-4-Mini Tutorial Covers Quantized LLM Workflows
This article walks through a notebook-based implementation using Microsoft’s Phi-4-mini-instruct model. It demonstrates 4-bit quantized inference, reasoning, tool use, retrieval-augmented generation, and LoRA fine-tuning in one pipeline.
MarkTechPost ↗