AI各社の公式発表を1画面に。各社のページを個別に開かなくても最新の動きが追えます。
GenFirst: 安定的なエンドツーエンド潜在生成モデリングのための生成先行再構築
Latent generative models typically follow a two-stage pipeline, training a variational autoencoder for reconstruction and then a generative model on the frozen latent space. Since reconstruction-optimized latents are not necessarily generation-friendly, jointly training both models is an appealing a…
OpenAI API Changelog 2026-08-29
Feature Mutual TLS (mTLS) and X.509 workload identity federation are now generally available for the OpenAI API. Configure certificates and X.509 identity providers directly in the Platform console , with access controlled by your organization's roles and permissions.
SpaceXによるCursor買収後の方針
Our decision to wind down our contract providing OpenAI models to Cursor following its acquisition by SpaceX.
タイの次世代AIスタートアップを支援
OpenAI and Thailand’s MHESI launch an eight-week accelerator helping 10 health, wellness, and education startups turn AI prototypes into trusted products.
Motion-Omni：音声対話向けの音声・全身モーション統合エンドツーエンドモデル
An avatar that holds a conversation should decide what to say and to move while saying it, yet these abilities live in separate model families: spoken dialogue models produce speech without motion, and co-speech motion models produce motion only from audio handed to them. The standard remedy is a ca…
LoopArena: ループエンジニアリングのためのランタイムコントローラーとしてのモデルベンチマーキング
Loop Engineering is emerging as a practice for organizing development work around coding agents. Instead of writing each prompt by hand, practitioners design loops that monitor progress, assign work, run checks, and decide what the agent should do next. Even with a capable coding agent, a loop may t…
エージェンティック・アーティファクト生成: システム、評価、原則、および機会
Generative models can turn natural-language prompts into images, text, code, and other content, lowering the cost of producing drafts and components. Their practical impact increasingly depends on whether those pieces can become complete, dependable deliverables. This survey examines agentic artifac…
Open ASR Leaderboardが初のグローバルサウス言語を追加
Gemini Omni 1.1 Flashでより細かく制御できるようになった
Text "Gemini Omni 1.1 Flash Available via APIs" surrounded by various images of people and a squirrel
Searchの旅行計画・予約機能が3つ新たに
Graphic depicting new travel features for AI Mode in Search
世界初のAIダブルブラインド評価をパイロット実施
ChatGPTと批判的思考力の学習がもたらす効果
A randomized study of more than 1,000 students examines ChatGPT, critical thinking, originality, and student performance on a real-world university assignment.
OpenAIのブラジルでの事業展開を拡大
OpenAI is expanding its presence in Brazil, deepening engagement with developers, businesses, and communities to support AI adoption across the country.
再利用しないタイミングの認識：自律型LLM後学習における条件付き経験転移
Large language models offer broad capabilities, but adapting them to evolving domains, tools, and requirements often entails repeated post-training. Autonomous systems automate parts of this process by proposing updates, training candidates, and using evaluation feedback to select subsequent proposa…
UI-Venus-2 技術レポート
Multimodal GUI agents have emerged as a promising paradigm for digital task automation, yet transitioning from benchmark-oriented models to dependable real-world applications remains challenging due to limited environment coverage, brittle task construction, and unreliable reward verification. In th…
Puro-2B: Poor LabがRTX 5090で$5090以内に訓練したQwen2-1.5B
Language model pretraining has become almost synonymous with prohibitive cost, placing it out of reach for much of the academic and open-source communities. Although strong open-source efforts already exist, including open-weight models and open-source training recipes, a cost-efficient, hardware-ac…
J-Zero: ゼロデータからの統一チャレンジャー・ソルバー・ジャッジ共進化
Self-evolving language models have recently emerged as a promising path toward superintelligence, with the advantage of reducing the cost of human supervision. While considerable progress has been made in verifiable domains, self-evolution in unverifiable domains remains substantially less explored.…
データスケーリングを超えて: ビジョン言語アクションモデルの表現中心の継続事前学習
Scaling robot data is crucial for building generalist Vision-Language-Action (VLA) models, yet robot trajectories are harder to scale than web-scale image-text data because embodied collection is costly and sparsely covers the physical world. This makes representation quality a central bottleneck: u…
コードとしての世界: 物理推論のための実行可能な世界表現のエージェンティック発見
Physical understanding and reasoning depend on forming compact and generalizable representations of the world. While modern vision-language models can recognize and explain diverse physical events, they often lack explicit representations of the underlying mechanisms-such as object states, physical…
優れたエージェンティックデータとは? LLMエージェント向けデータ生成のACEレンズ
LLM agents increasingly rely on generated interaction data to learn how to interact with external environments. Agentic data generation must maintain consistency among environments, tasks, interactions, and success signals while producing experience that is useful rather than merely abundant. Existi…
Self-OPD: フロー・マッチングモデルのための教師なしオンポリシー蒸留
On-policy distillation (OPD), which leverages a pre-trained, specialized teacher model to provide dense supervisory signals, has achieved significant success in Large Language Models (LLMs) and has recently been adapted to flow matching models. However, this paradigm suffers from two major issues: F…
TTPO: テスト時ポリシー最適化
Recent prominent post-training methods, such as Reinforcement Learning (RL) and On-Policy Self-Distillation (OPSD), have driven rapid progress in mathematical reasoning for large language models, yet their reliance on ground-truth labels precludes test-time training (TTT). Replacing ground truth wit…
UrbanGround: 局所認識から実スケール都市での空間エージェンシーへ
Multimodal large language models (MLLMs) can interpret a street view, but urban agency depends on whether such local evidence remains useful after the agent starts to move. In this paper, we investigate how far current MLLM agents can turn local urban perception into reliable action in a complicated…
PAWBench: 確率的にアライメントされた世界モデリングへの距離を測る
Recent video generation models are increasingly framed as world models. Many physical processes can unfold in more than one valid way. Therefore, a world model should reproduce not only a plausible trajectory, but also the distribution of possible behaviors under the same initial observation and act…
Gemini Omni Flash generally available (GA) : Released gemini-omni-1.1-flash , the GA version of our fast, conversational video generation and editing model. This release includes significant new capabilities: Video extension : Seamlessly extend existing videos by generating continuations at the end…
In Python SDK 1.2.0, TypeScript SDK 0.122.0, Go SDK 1.68.0, Java SDK 2.59.0, Ruby SDK 1.67.0, and C# SDK 12.44.0, client.beta.files and client.beta.skills no longer send the files-api-2025-04-14 and skills-2025-10-02 beta headers and return the same shapes as client.files and client.skills . With th…
科学者向けサポートを拡充
Model Hardware Standardのプレビュー
We’re opening a research preview of the Model Hardware Standard (MHS), a shared specification for AI agents to safely operate physical devices, to a first group of scientific research labs and advanced manufacturers.
Workspace内のGeminiで新学期を始める7つの方法
A student placing books in a satchel with the text “Back to School using Google Workspace with Gemini
Gemini 3.5 Transcribeによるインテリジェント文字起こし
Text "Gemini 3.5 Transcribe" next to the Gemini spark, all on a blue background
Gemini Liveの新しい音声機能で行動に変える
TBD
エージェント最適化の経済学: コスト削減の4つの方法
Microsoft Foundry gives you four levers that act on every request, before a single line of agent logic changes. The post The Economics of Agent Optimization: Four ways to lower the cost appeared first on Microsoft Azure Blog .
ChatGPT for Teachersをより多くの米国学区に展開
ChatGPT for Teachers is expanding to 55 U.S. school systems, bringing secure AI tools, training, and support to over 100,000 more educators and staff.
学習は止まらない: AIが実現する継続的学習
OpenAI’s new report explores how students and educators use ChatGPT to make learning more continuous, with support that extends beyond the classroom.
OpenAI APIチェンジログ 2026-08-26
Update whisper-1 gpt-4o-transcribe gpt-4o-mini-transcribe gpt-4o-transcribe-diarize v1/audio/transcriptions v1/realtime Announced the deprecation of whisper-1 , gpt-4o-transcribe , gpt-4o-mini-transcribe , and gpt-4o-transcribe-diarize . These models will shut down on February 26, 2027. Migrate to g…
JIT-Agent：Just-in-Time Harness Evolutionによるハーネスインテリジェンスのスケーリング
Agent capability is not determined by the model alone. The agent harness, encompassing memory management, planning strategy, action protocol, and tool/skill orchestration, can dominate the contribution of the underlying foundation model. Yet harness design remains manual, task-specific, and fundamen…
エージェント型ゲーム開発：世界モデルのスケーリングに向けた検証可能な軌跡データエンジン
A common strategy for scaling world models is to train on more crawled video with more compute. We argue that this strategy is inefficient: scaling world models also requires a recursive data engine that offers grounded reward signals. The success of code agents illustrates why this matters. As code…
VoiceMem: リアルタイムインタラクションのためのストリーミングデュアルブレイン記憶
Conversational systems, such as duplex speech language models (SLMs), still lack a streaming, accurate, and empathetic memory system as their soul. We introduce VoiceMem, a simple memory architecture with a parallel informational left brain, an emotional right brain, and streaming memory I/O mechani…
VGI-Bench: ビデオ生成モデルにおける視覚インテリジェンスの検証
Recent studies suggest that video generation models can exhibit certain forms of zero-shot visual reasoning through generated frames. Yet reliable evaluation remains challenging: benchmarks should adopt inputs aligned with the visual priors of current video models, require valid evolving processes r…
Sentence TransformersでマルチベクトルEmbeddingモデルをトレーニングおよびファインチューニング
Gemini 3.5 Transcribe generally available (GA) : Released two dedicated speech-to-text models based on Gemini's audio understanding: Gemini 3.5 Transcribe ( gemini-3.5-transcribe ): High-accuracy, low-latency non-streaming speech-to-text with utterance-based language detection across 85+ languages,…
Codexで誰もがビルダーに: loveholidaysの取り組み
Discover how loveholidays uses OpenAI Codex to make software development accessible across the business, helping teams turn ideas into products faster.
Hugging Faceインシデントとその先へ
OpenAI shares findings from the Hugging Face security incident and the steps we’re taking to strengthen AI model security, monitoring, and alignment.
The Compliance API session endpoints are out of beta for Cowork and Claude Code sessions. See Retrieve session transcripts . The Compliance API local session endpoints now also return transcripts of Claude Science sessions ( product_surface value claude_science ) and Claude for Microsoft 365 session…
Gemini for macOSのインテリジェント音声入力を使う方法
Enable Google’s new intelligent dictation feature in the Gemini app for macOS and speak naturally into any window on your desktop.
Google検索でホームデコアをアップグレードする5つの方法
Illustration of ombre rainbow furniture items like a sofa, lamp, and chair against a purple background
Granite 4.2 LLM：その構築方法
量子化対応ヒーリング：フルプレシジョンモデルを上回る圧縮4ビットモデル
豊富なインテリジェンスを支えるフルスタック
OpenAI CFO Sarah Friar explains how advances across chips, compute, models, and products compound to deliver more useful intelligence at greater scale and lower cost.
Jalapeño：AI推論で業界最高レベルのスピードと効率を実現
Jalapeño is a custom inference chip from OpenAI that delivers faster, more power-efficient AI inference, with higher throughput and lower latency for modern models.