2026-09-24 AI 日报

A living pixel-art station where real AI agents do real work. Local-first desktop agent harness - bring your own key, watch your crew actually run.

The cross-harness self-improving memory layer for AI agents.

A unified library of SOTA model optimization techniques like quantization, distillation, pruning, neural architecture search, speculative decoding, etc. It…

The open-source app everyone uses to manage agents at work

Google's open agentic orchestration runtime

Learn it. Build it. Ship it for others.

Multi-agent harness that runs Claude Code and Codex together as one system

The Office Harness for AI Agents — Spreadsheets, Docs, Slides, Canvas, Relational Tables, and PDF in one runtime.

Orca is the ADE for working with a fleet of parallel agents. Run any coding agent with your own subscription. Available on desktop, mobile and remote runtime.

Build an agent harness and control it end-to-end. Open-source SDK for production AI agents in Python & TypeScript - any model, any cloud.

Long-term conversational memory in multi-party settings requires more than retrieving relevant content from long-term conversations: it must distinguish who…

Spatial reasoning is essential for vision-language models (VLMs) to understand and act in the physical world. Reasoning in dynamic environments requires VLMs…

Advances in generative models have improved video fidelity, enabling long-horizon generation, interactive world modeling, and evolving visual environments.…

Evaluating world models requires assessing both the quality of the worlds they generate and their consistency and responsiveness under exploration,…

Reinforcement learning (RL) is vital for optimizing video generation models, with a robust reward model (RM) serving as the cornerstone. However, existing…

Agentic memory systems reuse past experience to improve future performance, yet most existing designs curate memory at write time: once a task is completed,…

Repository-level coding benchmarks have become the standard for evaluating coding agents, yet they inherently suffer from data leakage because they are built…

Robotic bin packing requires long-horizon sequential decision-making, as each object placement affects the available space for subsequent packing. Existing…

Vision-Language-Action models provide a strong foundation for general-purpose robot control, yet a vast majority of policies do not preserve and leverage…

Reinforcement learning has become a central component of large language model (LLM) post-training, yet token-level credit lacks a generally accepted…

Hey HN, I'm Jordan from Radix. Radix is a UI tool for programming agents. You prompt your agent to generate a workspace for a task you're working on and get…

@vasalex93 1. We will keep accelerating. Our AI efforts are only 3 years old, vs 6 and 10 years old for Anthropic and OpenAI. If our second derivative remains…

Situation today: GPT-6 Sol is like GPT-5.6 Terra GPT-6 Luna is like GPT-5.6 Asteroid https://t.co/D7OiG6KI8Q

Scroll to bottom for tldr In July, JetBrains reran the headline claims of two token-saving tools on real agent workloads. Caveman claimed 65% and measured…

I am 5th year and have no published research so far. All my papers have been consistently rejected from top tier AI conferences even though they got good…

Many a praise have been sung on Qwen-3.8, but here is mine. Qwen-3.8 and I had a rocky start, because it thinks so much. Watching it working is painful, so…

Hey everyone, Jovan from UkisAI here! Today, we are introducing Swift, a family of efficient reasoning LLMs based on Qwen, trained by penalizing tokens…

The user is frustrated — I rambled too much and didn't act. Let's just run the test suite and move on. No more forensics. One command, execute, then report.…

My paper advanced to the second round...but... Out of the papers I reviewed. One did not follow the AAAI template and was unblinded. My review was two lines.…

I have been thinking that the current Agentic Loop design of LLM Call -> Tool Call -> ... has been outdated. The arrival of Jev and other System One models…

We have a new acronym for you: SAFA. The Google, OpenAI and Anthropic AI safety body is starting to take shape, with a tentative name of the Standards…

Meet GPT-6 Sol and Luna, two models that bring frontier intelligence to everyday work with different balances of capability and cost.

We are excited to open-source Qwen-Image-2.1, an image model in the Qwen family that balances generation quality, inference efficiency, and cost.…

Gemini 3.8 Flash-Lite TTS and Gemini 3.8 Flash TTS are our most expressive audio models yet.

Google Cloud API Gateway now acts as a native remote Model Context Protocol (MCP) server, eliminating the need to build and maintain custom middleware to…

We’re announcing a new life sciences research group and laboratory at Anthropic. This post introduces the team behind this work and shares early results in…

Learn how GPT-6 improves prompt caching with higher cache hit rates, new diagnostics, explicit breakpoints, and controls that reduce latency and costs.

The MaxText team successfully reproduced AI2’s OLMo 3 7B language model from scratch on Google Cloud TPUs using JAX/XLA, precisely matching the original…