Google DeepMind introduced Gemini 4 Hydrogen on November 12 as a fast execution model for everyday coding and agentic workflows. Google says it scores 51 on the Artificial Analysis Intelligence Index at its default level, above Argon's 50, with flat pricing of $1.50 per million input tokens and $7.50 per million output tokens. All figures come from Google.
Google released the Gemini Decisions API, powered by Gemini Decision 1. It does not write text. It scores closed sets of options in one forward pass and returns probabilities and confidence. Under its benchmark with inputs up to 16K tokens and 1–100 questions per request, Google claims 150–250 ms latency. Input costs $0.10 per million tokens, and output is free.
Ars Technica, citing Reuters, reports that TSMC's board approved up to $500 million in OpenAI preferred stock and $500 million in Anthropic common shares at the expected IPO price. TSMC has posted six straight record-profit quarters, with over $23 billion in July to September, while analysts warn that circular financing blurs real demand.
Cloudflare open-sourced Kitesurf on GitHub under Apache 2.0. It is a stateless browser for AI agents that runs entirely on Workers. It parses HTML, applies CSS, runs JavaScript and renders screenshots without a browser container, and it supports CDP so Puppeteer can connect.
AWS released AWS Emulator, a local emulator that runs on your machine. The AWS CLI, SDKs and coding agents call it over a local HTTPS endpoint with the same AWS APIs, so application code stays unchanged. AWS says there is no extra service fee, and the source is on GitHub.
Meta's Reality Labs ported Cuttlefish, the AOSP virtual-device framework, to macOS and Apple silicon. The work touches QEMU's vhost-user path, HVF memory tracking, the native Cocoa display, and build and signing. It ships as a QEMU fork and a patch set.
IBM introduced LLMON on the Hugging Face blog, a markup language that separates instructions from data, with 3B and 8B Granite 4.1 preview models.
Vercel says Daemons can send OpenTelemetry traces to Vercel Tracing. In a default Node.js daemon, call registerOTel before worker initialization, and @vercel/otel 2.3.0 or later exports the spans the worker records.