# AI Dev News (Marktechpost.com)

> Empowering AI builders with the latest news in AI, machine learning, enterprise platforms, and software development.

This file provides information about AI Dev News (Marktechpost.com) to help large language models understand and reference this publication's content.

## Posts
- [📡 Kimi K3 Drops at 2.8T; GPT-Red Crushes Human Red-Teamers ](https://www.aidevsignals.com/p/kimi-k3-drops-at-2-8t-gpt-red-crushes-human-red-teamers-7976): A 2.8-trillion-parameter open model, an embedding checkpoint that just topped the retrieval leaderboard, an AI that out-hacks human red teamers 6-to-1, and xAI open-sourcing its coding agent. The frontier is moving in every direction at once — here's what shipped.
- [📡 Mira Murati Ships Her First Frontier Open Model — and a new 27B Intelligence Fits on a Phone](https://www.aidevsignals.com/p/mira-murati-ships-her-first-frontier-open-model-and-a-new-27b-intelligence-fits-on-a-phone): Thinking Machines Lab finally shipped an open model — Inkling, a 975B open-weights MoE. PrismML squeezed a 27B model down to 3.9GB so it runs on an iPhone. Google put real ML inference in the browser. And Europe got a serious open bilingual foundation model.
- [📡 Mistral Steers Robots With One Camera and Prime Intellect rebuilt its RL environment stack](https://www.aidevsignals.com/p/mistral-steers-robots-with-one-camera-and-prime-intellect-rebuilt-its-rl-environment-stack): Mistral had a loud week — an 8B robot navigator that runs on a single RGB camera, plus a head-to-head benchmark pitting its own coding agent against Claude Code, Cursor, and Codex. Prime Intellect rebuilt its RL environment stack. And two sharp open-source dev tools shipped: one for domains, one for docs.
- [📡 The Week NeuroVFM Reads Raw Hospital Scans; Stanford's TRACE Fixes Broken Agents](https://www.aidevsignals.com/p/the-week-neurovfm-reads-raw-hospital-scans-stanford-s-trace-fixes-broken-agents): A neuroimaging foundation model trained on 5.24M raw hospital scans. Mira Murati's Thinking Machines Lab making the technical case for AI you can actually own. Stanford turning agent failures into training signal.
- [📡 The Week of Physical AI: Ant Group Shipped 4 Embodied Models](https://www.aidevsignals.com/p/the-week-of-physical-ai-ant-group-shipped-4-embodied-models): In five days, Ant Group's Robbyant shipped four open releases that together cover the entire pipeline of physical intelligence — from how a robot sees, to how it acts, to how it imagines the world, to how it controls its own body in real time. 
- [📡 The Week Meta Brings Muse Spark 1.1 and Ant Group Leads in Physical AI](https://www.aidevsignals.com/p/the-week-meta-brings-muse-spark-1-1-and-ant-group-leads-in-physical-ai): Robbyant open-sourced a world model that stays coherent for a full hour and runs on a single GPU. Meta shipped its first paid model and undercut OpenAI and Anthropic on price. NVIDIA squeezed 2x server throughput out of a compressed MoE. And Ollama raised $65M to keep local AI local.
- [📡 The Week Grok 4.5 Ships and Robbyant Releases LingBot-VLA 2.0](https://www.aidevsignals.com/p/the-week-grok-4-5-ships-and-robbyant-releases-lingbot-vla-2-0): SpaceXAI and Cursor shipped a 1.5-trillion-parameter coding model. OpenAI's voice models can now listen and talk at the same time. GPT-5.6 finally clears government review for a full Thursday launch. And Ant Group's Robbyant just taught one 6B model to control 20 different robots.
- [📡 The Week Ant Group Solves Spatial Perception at 1B Params, Tencent Ships a 295B Open Model](https://www.aidevsignals.com/p/the-week-ant-group-solves-spatial-perception-at-1b-params-tencent-ships-a-295b-open-model): Ant Group's Robbyant open-sourced a 1B vision model that beats models 7x its size at dense spatial perception. Tencent dropped a 295B MoE under Apache 2.0. NVIDIA unified audio and text in one decoder without losing IQ. And Liquid AI cut "doom loop" rates in reasoning models by 90% with one training trick. 
- [The Week Fable 5 Comes Back, China Closes the Gap, and Proof Assistants Get an Agent](https://www.aidevsignals.com/p/the-week-fable-5-comes-back-china-closes-the-gap-and-proof-assistants-get-an-agent): Anthropic's Fable 5 is back online after 18 days dark. Zhipu's GLM 5.2 landed within 1 point of Opus 4.8 at a fifth of the cost. Mistral solved 587 of 672 Putnam problems with an open model. And three separate teams shipped agents that can read — your browser, your webpage, and your Fitbit. Here's everything.
- [The Week Governments Started Gating Frontier AI and Open-Source Filled the Vacuum](https://www.aidevsignals.com/p/the-week-governments-started-gating-frontier-ai-and-open-source-filled-the-vacuum): GPT-5.6 landed — then got restricted to 20 trusted partners at the White House's request. Meanwhile: DeepSeek made its own models 85% faster without a new GPU. Liquid AI shipped a 230M model that runs on a mobile NPU. Meta open-sourced a design system agents can actually read. Here's everything.
- [The Week Apple Released Container & Baidu Released Unlimited OCR](https://www.aidevsignals.com/p/the-week-apple-released-container-baidu-released-unlimited-ocr): Apple dropped a containerization tool that runs Linux as lightweight VMs on Apple Silicon — no Docker, pure Swift. DeepReinforce open-sourced a coding model that matches closed frontier systems. 
- [📡 AI INTELLIGENCE BRIEF: The Week PDFs Got a Schema, Kernels Got Faster, and Grok Went Autonomous](https://www.aidevsignals.com/p/ai-intelligence-brief-the-week-pdfs-got-a-schema-kernels-got-faster-and-grok-went-autonomous): Datalab ships a 9B PDF-to-JSON extractor. Prime Intellect trains trillion-parameter MoE with RL. Sakana routes tasks across frontier LLMs through one API. xAI drops /goal for long-running autonomous coding. GPT-5.6 previewed, Noam Shazeer joins OpenAI. Here's everything from the last 4 days.