Nvidia Halts H200 Sales to China: What Actually Happened
Nvidia says it has stopped shipping licensed Hopper (H200) data center chips to China. It’s not “all AI chips,” and Nvidia gave no reason. Here’s what’s confirmed and what isn’t.
Alibaba Qwen4-Max Explained: 2.4T Parameters, 1M Context, Real Price
Alibaba launched Qwen4-Max on November 16, 2026: 2.4T parameters, 71B active, 1M context, native vision. Here’s the real price, what the 60% cost claim means, and what’s still unproven.
DeepSeek DeepInfer Explained: Is It Really Making GPUs 3x More Valuable?
DeepSeek open-sourced DeepInfer, its in-house inference engine. The “3x more valuable GPUs” headline is a stretch. Here’s what the release really is, what it runs on, and what’s missing.
Mirendil Raises $1 Billion Series A: What the AI Agent Startup Actually Builds
Mirendil, founded by ex-Anthropic researchers, announced a $1 billion Series A on November 12, 2026. Here’s what it builds, who backed it, and the $5 billion valuation claim nobody has confirmed.
DeepSeek Puts Its App Features Into the API: Search, Files and Voice. Here’s What It Really Costs.
DeepSeek now lets developers use its app’s web search, link reading, file parsing and safety checks through the API. Here’s how it works, what it costs, and the limits nobody mentions.
Gemini 4 Hydrogen Explained: Google’s Fast, Cheap Model for Agents and Search
Google’s Gemini 4 Hydrogen is a fast, low-cost model for agent work, now in Search and the Gemini app. Here’s the price, the benchmarks, how Orchestrator works, and what’s only Google’s claim.
AlphaGenome 2 Explained: What DeepMind’s New DNA Model Can Do, and What It Can’t
Google DeepMind’s AlphaGenome 2 predicts how DNA variants act in specific cell types. Here’s what’s new, the benchmark numbers, who can use it, and why it can’t diagnose disease.
Gemini 4 Carbon Is Now Available. Here’s the Price, the Access and the Catch.
Google opened Gemini 4 Carbon to paying developers and AI Ultra subscribers on November 10. Here’s the price, who gets it, how it compares with Claude Fable 5.2, and which claims are Google’s own.
Google’s New TPU 8tx and 8i Max Explained: Two Chips Aimed at Nvidia, But Not the Way You’ve Read
Google unveiled two specialized TPUs: 8tx for confidential computing and 8i Max for long-context AI. Here’s what’s confirmed, what’s Google’s own claim, and what it means for Nvidia.
Receive the latest articles in your inbox
Signup for Newsletter