2026-11-09 AI 日报

The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode,…

DeepSeek Harness: Everything is a Plugin.

Turn any codebase, with its docs, SQL schemas, configs, and PDFs, into a queryable knowledge graph. A /graphify skill for Claude Code, Cursor, Codex, and…

Ralph is an autonomous AI agent loop that runs repeatedly until all PRD items are complete.

Repo for the Complete Agentic AI Engineering Course

llama.cpp fork for significantly improved performance on Ampere generation Nvidia cards (especially RTX 3090 / 3090 Ti): example: 110+ tok/s over a 100K-token…

Official inference framework for 1-bit LLMs

AI Kanban & Development Environment. Orchestrate multiple agents, review changes, open PRs. Multi-provider, self-hostable, no telemetry.

A series of Jupyter notebooks that walk you through the fundamentals of Machine Learning and Deep Learning in Python using Scikit-Learn, PyTorch, and Hugging…

12 weeks, 26 lessons, 52 quizzes, classic Machine Learning for all

We identify a key limitation of existing feed-forward 3DGS methods: Gaussian attributes are typically predicted in a radial-distance-normalized space that…

Streaming speech recognition is increasingly consumed by machines, such as translators and voice agents, that act on each token as it appears. A token that is…

Multi-teacher on-policy distillation consolidates expert capabilities by supervising student-generated trajectories, whereas task arithmetic offers a…

Generative video models are increasingly used as visual world simulators, yet their primary strength lies in rendering plausible pixels rather than reasoning…

We introduce LongBanana, to our knowledge the first expert-designed and expert-verified benchmark for long-context multi-reference image synthesis in the age…

Agent harness evolution enables recursive self-improvement (RSI) without updating model weights. We examine whether higher benchmark scores yield reliable,…

Recent years have witnessed remarkable progress in video generation in terms of visual quality and temporal consistency. However, generating videos with…

Relative-position attention mechanisms offer compelling alternatives to RoPE across various transformer architectures, delivering superior performance in…

Solving long-horizon agentic tasks requires language agents to interact with their environments over multiple steps, and reliable performance often comes from…

Weight-Space Programming (WSP) enables reasoning agents to directly inspect and modify neural network parameters, offering an alternative to gradient-based…

The agent running on a remote machine can authenticate using your local SSH credentials. That means it can push to your private repos on GitHub, for example.…

Yet another new Gemini model: Google released a new internal checkpoint called Krypton. The results so far are really impressive and even better than Carbon.…

I've been experimenting with having Fable 5.1 write every weight of a neural network straight from a written spec. Staring from zero, with no training data…

FastAPI's official AI development guide has a useful detail: its agent skill ships with the Python package, with guidance for the version installed in your…

Are there any (insider) infos to upcoming Gemma Models? Gemma 4 is for my usecase really good and I am looking forward if they will release any new even…

Just found out that Claude subscriptions are cheaper if you purchase them through the Apple App Store. For example: Claude Pro (Website): ₹2,400/month Claude…

Guys, I don't know if it's just me, but over the last three days I feel like OpenAI's models have suddenly gotten dumber... I'm developing a WordPress plugin.…

Prefixwise tells you why your LLM calls miss the prompt cache, and what each miss costs you. Put it between your agent and the model, or inside vLLM itself,…

source https://x.com/chetaslua/status/2119695384843108458

Who else keeps seeing this ad for Local AI with a middle grade gaming case, no GPU and no RAM. It makes me so mad 😅

Gemini Omni 1.2 Flash brings native 15-second video generation and editing, stronger consistency, and 30% lower pricing.

Pathways is a single-controller distributed training system designed to coordinate large-scale model training across multiple accelerator groups by…

Released GPT-Image-2.6 Sunburst and GPT-Image-2.6 Flare for image generation and editing, with adjustable complexity, SVG generation, and animated GIF…

Use GPT-6-Cyber to investigate vulnerabilities, reproduce them in an authorized test environment, and verify fixes. See Daybreak setup before sending your…

Images 2.6 brings faster image generation, GIF creation, more precise editing, and new API controls across ChatGPT, Codex, and the API.

Today, we're introducing spaces, an interactive way to build a more customized experience in the Muse app. The post Design the Muse App, Just for You appeared…

Turn your sketches into functional apps using Gemini. Upload your drawings to the Canvas tool to generate and refine code instantly. Try it now.

Eligible Enterprise organizations can now tag projects with the codes they already use, like a charge code or client-matter number, and see most spend in…

You can now sign a member of your Claude Enterprise (claude.ai) organization out of Claude everywhere with the Admin API. This capability is in beta. See the…

You can now set up an IP allowlist for your organization in the Claude Console, in beta. Requests to covered services from addresses not on your list are…