Huuush · Knowledge
AI models, product launches, capabilities, pricing, and performance.
98 articles
AI & Services
The difference between the sidebar extension that puts Claude inside Google Docs, Sheets, and Slides and the Google connector that creates and edits files from a Claude conversation, plus plans and install requirements.
Oct 6, 2026
AI & Services
Nano Banana 2.1 supports image generation and editing, up to 14 reference images, and search grounding. Gemini API Standard and Batch pricing and cost by resolution are covered too.
Oct 6, 2026
AI & Services
How to coordinate several cloud threads in Claude Code Projects from mobile, and the Remote Control conditions needed for tasks that use local files and Chrome.
Oct 6, 2026
AI & Services
Preview Markdown in Google Drive and edit, comment, and collaborate in Docs without converting the file format. Here is how it works and where it's rolling out.
Oct 6, 2026
AI & Services
Cloudflare Web Search API search providers and pricing, BYOK setup, and REST and Workers call paths, plus the difference between search provider ZDR and AI Gateway logs.
Oct 6, 2026
AI & Services
How OpenAI's textGrain text watermark works, how it applies to ChatGPT and Codex in the EU and as an opt-in setting in the API, and the detector access and detection conditions.
Oct 6, 2026
AI & Services
How NVIDIA SkillSpector checks AI agent Skills before installation, the difference between static analysis and optional LLM analysis, and what data gets sent out.
Oct 5, 2026
AI & Services
Matt Pocock's chief-of-staff Skill is an in-progress beta that coordinates subagents and schedules under a long-term goal and guides you to split strategy and priorities from tactical execution.
Oct 5, 2026
AI & Services
Cactus Compute's Whistle is an on-device speech recognition model that runs on a CPU from a single 16.9 MB model file. It supports 7 languages and 17 deployment targets, and the browser demo processes audio on your device after the first model download.
Oct 5, 2026
AI & Services
Cloudflare's Clef and Clef-Flash are open-weight decision models that return per-option probabilities. What they offer: a Jev-compatible API, local execution, and RL fine-tuning.
Oct 3, 2026
AI & Services
Andrej Karpathy suggests ways to turn an LLM's answer into forms that are easier to understand. It starts with ASD-STE100-style concise sentences, moves to diagrams, interactive HTML, and custom explainer videos, and ends with the idea of one-off custom software.
Oct 3, 2026
AI & Services
DeepSeek Harness v0.2 preview arrives as a macOS and Windows desktop app that handles documents, spreadsheets, PDFs, and code in a single conversation.
Oct 2, 2026