AN IN-DEPTH REPOSITORY
Every model, paper, media release, agent and tool from the month—in one source-linked archive.
COMPLETE MONTHLY ARCHIVE
The frontier shipped: faster model families, natural voice, embodied navigation and the infrastructure needed to put agents into real work.
THE COMPLETE MONTHLY INDEX
Every linked paper, product release, research project and article has been checked directly. Cards use the publication's own description, imagery and published date wherever the source supplied them.
The AI Search helps discover the wider field, while every available card opens the original publication. Dates come only from publication metadata or official source APIs—never from a video. A checked date records our review, not the release. Older undated entries show their archive month.
HUGGINGFACE.CODeepSeek-V4-Flash-0731
DeepSeek’s official V4 Flash release substantially improves agentic performance over the preview while retaining its speculative-decoding module and supporting low, high and max reasoning effort.
MINIMAX.IOMiniMax H3: An Open Model Breaking the Boundaries Between Tasks and Modalities - MiniMax Research
Today, we're officially launching MiniMax H3, a general-purpose omni-modal generation model. H3 can jointly understand multimodal contexts spanning text, images, video, and audio. It generates video with native stereo audio at up to 2K resolution and 15…
DEEPMIND.GOOGLEGemini Robotics 2 brings whole body intelligence to robots
From feet to fingertips — we are teaching robots intelligent whole-body control, fine dexterity, and teamwork to complete a broad range of complex tasks.
OPENAI.COMAdvancing the price-performance frontier with GPT-5.6
OpenAI cut GPT-5.6 Luna pricing by 80% and Terra pricing by 20%, while introducing a Fast API mode for higher-throughput GPT-5.6 Sol workloads.
GITHUB.COMGitHub - Pan-Chera/Multi-Agent-CAD: MAC (Multi-Agent CAD): A decoupled multi-agent framework for text-to-CAD generation via constrained test-time compute
MAC (Multi-Agent CAD): A decoupled multi-agent framework for text-to-CAD generation via constrained test-time compute - Pan-Chera/Multi-Agent-CAD
RESEARCH.GOOGLEScience One Framework: A verifiable autonomous research framework via Chain-of-Evidence
Google Research introduced an autonomous research prototype that builds an auditable chain of evidence as it works, paired with CoE Audit for checking the integrity of AI-generated papers.
BLOG.GOOGLEGemini for macOS adds new natural language capabilities
A look at how the Gemini app for macOS now lets you speak naturally to get clean transcriptions, edits, and summaries just using your voice.
Lyria 3.5 — Model Card
Google's latest music-generation model improves musicality, vocals, lyrics, audio fidelity and prompt control while producing cohesive tracks as long as three minutes in Flow Music.
ARXIV.ORGQwen-Audio-3.0-Gen-Preview Technical Report
Qwen’s technical report presents a unified audio-generation model spanning speech, music and general sound, with controllable generation and a shared multimodal architecture.
Visual prompt engineering for video models
Google DeepMind found that automatically editing a task image before inference can improve video-model reasoning more than text-only prompt engineering or extra test-time compute.
HUGGINGFACE.CONVIDIA Cosmos-H-Dreams: Bringing Real-Time Generative Simulation to Surgical Robotics
NVIDIA released a real-time, action-conditioned surgical world model that lets operators and learned robot policies act inside generated surgical scenes and observe the results interactively.
HUGGINGFACE.COThinking Machines released a 276B-total, 12B-active open multimodal model that accepts text, images and audio for coding, tool use, conversational and retrieval workflows.
ANTHROPIC.COMIntroducing Claude Opus 5
Opus 5 is a step change improvement for the Opus tier powering long-running agents while delivering improvements in coding and professional work.
BFL.AIFLUX 3: One Multi-Modal Model | Black Forest Labs
One multi-modal model for Image, Video, Audio and Action-Prediction. Creations are truer to life in every kind of style.
HUGGINGFACE.CObaseten/GLM-5.2-Vision-NVFP4 · Hugging Face
GLM 5.2 Vision NVFP4 is published on Hugging Face. Open the model or project card for the author’s documentation, files, license and usage details.
LEARN.CHATGPT.COMChatGPT Voice | ChatGPT Learn
Try voice in Chat, Work, and Codex in the ChatGPT desktop app.
OPENAI.COMHealth in ChatGPT links to its original tools page, but the publisher did not expose a readable preview. Open the source directly for the full release.
HOMIE Demo Page
Human-object Centric Video Personalization via Multimodal Cognition Enhancement
POOLSIDE.AIIntroducing Laguna S 2.1
Today we’re releasing Laguna S 2.1, a significant step forward in our development of models that pursue longer horizon work and make effective use of reasoning.
HUGGINGFACE.COmicrosoft/Mage-Flow · Hugging Face
Mage-Flow is published on Hugging Face. Open the model or project card for the author’s documentation, files, license and usage details.
MODELSCOPE.AINanbeige4.2-3B
ModelScope——A one-stop service that brings together the most advanced machine learning models from various fields, offering model exploration experience, inference, training, deployment, and application.
OPENAI.COMOpenAI–Hugging Face security incident links to its original research page, but the publisher did not expose a readable preview. Open the source directly for the full release.
NEXT-STATE.GITHUB.IOHow to train a frontier-level world model
How we trained and open-sourced a frontier-level world model — the lessons, failures, and fixes — with a live, playable demo running on Reactor.
Qwen Studio
Qwen Studio offers comprehensive functionality spanning chatbot, image and video understanding, image generation, document processing, web search integration, tool utilization, and artifacts.