気になった記事やポストを個人的なメモと共に記録しています。
Reinforcement learning towards broadly and persistently beneficial models
Claude Code now supports artifacts
The White House Wants Anthropic to Block All Jailbreaks. That May Not Be Possible
HappyOyster 1.0 is now live!
GLM-5.2: Built for Long-Horizon Tasks
Predicting model behavior before release by simulating deployment
Agentic coding and persistent returns to expertise
ポケモンカードゲーム AI Battle Challenge
Meet Dreamina Seedance 2.0 Mini
Sakana AI、初の商用プロダクト「Sakana Marlin」を提供開始
Meet Kimi K2.7 Code HighSpeed!
アメリカ政府が Anthropic に対し、Claude Fable 5 と Mythos 5 へのアメリカ人以外のアクセス停止を指示
Statement on the US government directive to suspend access to Fable 5 and Mythos 5
Policy on the AI Exponential
New in Claude Managed Agents: run agents on a schedule and store environment variables in vaults
DiffusionGemma: 4x faster text generation
Latent Spatial Memory for Video World Models
Designing loops with Fable 5
Introducing North Mini Code: Cohere’s first model for developers
NVIDIA CUDA 13.3 Enhances GPU Development with Tile Programming in C++, Compiler Autotuning, and Python Updates
Claude Fable 5 and Claude Mythos 5
Fluid, natural voice translation with Gemini 3.5 Live Translate
Introducing the Google Colab CLI
FLUX.2 is now on device: ASUS ProArt laptops now support Klein models
Loop Engineering
A broccoli farmer in northern Japan shares his chats
Making Claude a chemist
Google ColabでVision LLMを作る
Codex SDK
Your largest local models, now in your pocket
Magenta RealTime 2 (Apps & Plugins)
Dreaming: Better memory for a more helpful ChatGPT
CUDA Programming Guide Part 1
In Support of Mandatory Nucleic Acid Synthesis Screening and Recordkeeping
NVIDIA Nemotron 3 Ultra Powers Faster, More Efficient Reasoning for Long-Running Agents
Introducing new capabilities to GPT‑Rosalind
Introducing Gemma 4 12B: a unified, encoder-free multimodal model
The Agent That Grows With You
NVIDIA Partners With Microsoft on Unified Stack for Agentic AI Deployment, From Windows Devices to Cloud to Local
All Stations Go: Developers Around the World Power Up NVIDIA DGX Station
NVIDIA GTC Taipei at COMPUTEX: Live Updates on What’s Next in AI
Introducing Majorana 2
Windsurf is now Devin Desktop
Codex for every role, tool, and workflow
OpenAI frontier models and Codex are now available on AWS
Expanding Project Glasswing
Anthropic confidentially submits draft S-1 to the SEC
Qwen3.7-Plus: Multimodal Agent Intelligence
Introducing MiniMax M3: The First Open-Weights Model to Combine Three Frontier Capabilities
LocateAnything - Fast and High-Quality Vision-Language Grounding with Parallel Box Decoding
Introducing dynamic workflows in Claude Code
Introducing Claude Opus 4.8
Zero Trust for AI agents
Coralboard
How we contain Claude across products
An OpenAI model has disproved a central conjecture in discrete geometry
Google Antigravity SDK
Google Antigravity CLI
Introducing Google Antigravity 2.0
Gemini 3.5: frontier intelligence with action
OpenAI Guaranteed Capacity
Use Grok in OpenClaw
New in Claude Managed Agents: self-hosted sandboxes and MCP tunnels
Composer 2.5 の紹介
Connect Grok to Hermes Agent
A new personal finance experience in ChatGPT
Claude Code を大規模コードベースで使うベストプラクティス
Grok Build Beta
Work with Codex from anywhere
Building a safe, effective sandbox to enable Codex on Windows
クラウドエージェントの開発環境
jina-embeddings-v5-omni: Embeddings for Text, Image, Audio and Video
Introducing Perceptron Mk1
Bringing the best of Gemini in Chrome to Android
Introducing Googlebook, designed for Gemini Intelligence
A smarter, more proactive Android with Gemini Intelligence
Introducing PROWL: Learning Through Discovery
Microsoft Teams で Cursor が利用可能に
OpenAI launches the OpenAI Deployment Company to help businesses build around intelligence
TransformerのSelf AttentionのQKVを直感的に解説する
EMO: Pretraining mixture of experts for emergent modularity
Investigating the consequences of accidentally grading CoT during RL
Running Codex safely at OpenAI
Coding Agent比較用の独自のベンチマーク、Harness Benchを作ってみた話
Introducing Trusted Contact in ChatGPT
Advancing voice intelligence with new models in the API
Natural Language Autoencoders: Turning Claude’s thoughts into text
AlphaEvolve: How our Gemini-powered coding agent is scaling impact across fields
Multi-Teacher On-Policy Distillation: A New Post-Training Primitive
New in Claude Managed Agents: dreaming, outcomes, and multiagent orchestration
Agents for financial services
GPT‑5.5 Instant: smarter, clearer, and more personalized
How SSA Makes Long Context Practical
Introducing SubQ: The First Fully Subquadratic LLM
How OpenAI delivers low-latency voice AI at scale
Claude Code入りのDockerイメージをDevContainerで動かす
Codex pets を試す
Introducing Advanced Account Security
Agents can now create Cloudflare accounts, buy domains, and deploy
Introducing Moonlake's 3D Agent: Computer Use Capabilities for World Modeling
WebSockets の Responses API におけるエージェントワークフローの効率化
Cursor SDK でプログラム可能なエージェントを構築する
Warp is now open-source
OpenAI models, Codex, and Managed Agents come to AWS
SpAItial is pioneering Physically-grounded World Models
「GitHub Copilot」従量課金に トークン消費量ベースで請求へ “定額使い放題”時代の終わりか
Qwen-Image-2.0-pro
ガバメントAI「源内」をオープンソースとして公開します
DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence
Grok Voice Think Fast 1.0
Introducing GPT‑5.5
Introducing Gemini Enterprise Agent Platform, powering the next wave of agents
Our eighth generation TPUs: two chips for the agentic era
ChatGPT にワークスペースエージェントが登場
Deep Research Max: a step change for autonomous research agents
Cursor、モデル学習でSpaceXと提携
ChatGPT Images 2.0 が登場
ゼロから作る日本語 LLM - GPT-2 の推論・学習の可視化から Modal での事前学習まで
Kimi K2.6: Advancing Open-Source Coding
Chronicle - Build Codex memories from recent screen context
EDINET-Bench: 有価証券報告書を用いた日本語金融ベンチマークの公開
Vercel April 2026 security incident
Introducing Claude Design by Anthropic Labs
Best practices for using Claude Opus 4.7 with Claude Code
Qwen3.6-35B-A3B: Agentic Coding Power, Now Open to All
Accelerating the cyber defense ecosystem that protects us all
Introducing GPT‑Rosalind for life sciences research
Introducing Ternary Bonsai: Top Intelligence at 1.58 Bits
Codex for (almost) everything
Introducing Claude Opus 4.7
クラウド版 Dataform のワークフローを JSON で管理する GitHub Action を作った
apply-dataform-workflows
The next evolution of the Agents SDK
HY-World 2.0: A Multi-Modal World Model for Reconstructing, Generating, and Simulating 3D Worlds
Gemini 3.1 Flash TTS: the next generation of expressive AI speech
Gemini アプリが Mac に登場
Automated Alignment Researchers: Using large language models to scale scalable oversight
Trusted access for the next era of cyber defense
ソフトバンクが国産AIの新会社設立、NECやホンダなど8社出資
Introducing Muse Spark: Scaling Towards Personal Superintelligence
Scaling Managed Agents: Decoupling the brain from the hands
System Card: Claude Mythos Preview
Project Glasswing
Gemma 4: Byte for byte, the most capable open models
Transform your headphones into a live personal translator on iOS
リアルタイムRLでComposerを改善する
Build real-time conversational agents with Gemini 3.1 Flash Live
A foundation model of vision, audition, and language for in-silico neuroscience
動画生成を蒸留で27倍速くした話
自社インフラストラクチャでクラウドエージェントを実行
TurboQuant: Redefining AI efficiency with extreme compression
Claude Code auto mode: a safer way to skip permissions
Harness design for long-running application development
Arm expands compute platform to silicon products in historic company first
最大規模のオープン基盤モデルを各国仕様へ適応させる事後学習技術を開発
Announcing TypeScript 6.0
PyTorch 2.11 Release Blog
Put Claude to work on your computer
Long-running Claude for scientific computing
Designing delightful frontends with GPT-5.4
Claude Codeの使用率がステータスラインに表示できるようになったので表示用のスクリプトを作った話
OpenAI to acquire Astral
GPT‑5.4 mini と nano が登場
日本語の手書きメモを書き起こせるOCRを探すために21モデルを片っ端から試した話
音楽の生成・編集が可能な高性能ローカル音楽生成AI【ACE-Step-1.5】から音楽生成AIの仕組みを完全に理解する
自宅で動くLLMをどこからでも呼び出せる「LM Link」、Tailscale×LM Studio連携で実現
MicroGPT explained interactively
Coding Agent時代の開発ワークフローについてのまとめ
Beyond the Limit: Introduce Mixedbread Wholembed v3
From model to agent: Equipping the Responses API with a computer environment
Four MTIA Chips in Two Years: Scaling AI Experiences for Billions
Gemini Embedding 2: Our first natively multimodal embedding model
autoresearch
Claude Code / Codex ユーザーのための誰でもわかるHarness Engineeringベストプラクティス
openai/symphony
Voice mode in Claude Code
新しい時代の開発と組織について
Our agreement with the Department of War
Manage Claude's memory
Statement from Dario Amodei on our discussions with the Department of War
Claude Code Remote Control
The path to ubiquitous AI
GPT‑5.3‑Codex‑Spark のご紹介
Unrolling the Codex agent loop