OpenAI just released Privacy Filter, an open-weight model designed to detect and redact personally identifiable information in text. Built with 1.5B total and 50M active parameters, it supports up to 128,000 tokens of context and runs locally so sensitive data never leaves your device. It achieves a 97.43% F1 score across eight privacy categories.
Open CoDesign is an open-source, desktop alternative to tools like Claude Design and Figma AI. It lets you bring your own model keys—from Claude and GPT to local Ollama—to turn prompts and reference files into interactive UI prototypes, slides, and exportable HTML.
A new open-source algorithm called RotorQuant just outperformed Google's TurboQuant across every metric. By swapping heavy global transforms for compact Clifford rotators, it hits 10x KV cache compression while boosting prefill speed by 5.3x and decoding by 28%. Why it matters: massive 128K context windows that previously required multiple enterprise GPUs can now fit onto a single 24GB card.
Overworld just dropped Waypoint-1.5, an open-source world model that lets you explore and interact with generative environments in real time. Running locally on everyday GPUs like an RTX 3070 at up to 720p and 60 fps, it effectively shrinks what once required a GPU cluster down to a home gaming rig.
ACE Music just released ACE-Step 1.5 XL, an open-source 4B DiT music generator capable of producing full vocal tracks on a local PC in under a minute. The distilled version runs in just 8 steps and rivals leading commercial tools, though smooth local execution requires at least 12GB of VRAM.
Researchers just open-sourced INSPATIO-WORLD, a real-time simulator that transforms monocular video into an interactive 3D scene you can actually walk through. Instead of guessing raw pixels, it maintains a persistent internal representation, hitting 10 fps on a single consumer RTX 4090. Why it matters: interactive dynamic world models are finally running on everyday GPUs.
Factory.ai just launched a native desktop app for macOS and Windows, letting users run multiple AI Droids simultaneously with dedicated contexts and persistent environments. The app can control local tools like VS Code and browser tabs, while supporting air-gapped local models via Ollama or vLLM. Early enterprise data shows users juggling both CLI and desktop complete 4.6x more sessions.
Google just released Google AI Edge Eloquent on iOS, a free offline-first dictation app built to take on tools like Wispr Flow and SuperWhisper. Powered by on-device Gemma models, it transcribes speech in real time while stripping out filler words like "um" and "ah" and polishing messy thoughts into bullet points or formal prose. An optional cloud mode taps Gemini for heavy cleanup, with an iOS keyboard extension coming soon.
Ai2 just open-sourced WildDet3D, a lightweight vision model that estimates an object's real-world 3D position, size, and orientation from a single RGB photo or camera stream. It supports text prompts, point clicks, or 2D boxes, and can even run in real time on an iPhone. Why it matters: It gives robots and AR apps full spatial intelligence without requiring complex multi-camera rigs.
Ben Sigman and actress Milla Jovovich just released MemPalace, an open-source AI memory system built with Claude that achieved the first perfect 500/500 score on LongMemEval. Instead of sending data to the cloud, it runs locally, organizing memories into a palace architecture and compressing user context down to ~120 tokens.
Google just launched Gemma 4, its most intelligent open model lineup yet, released under a commercially permissive Apache 2.0 license. Spanning four sizes from on-device E2B to a 31B dense flagship, the models handle vision, speech, up to 256K context, and agentic workflows, with its 26B MoE outranking models 20 times its size. It basically delivers frontier-level reasoning to local hardware with minimal overhead.
LongCat just launched LongCat Audio Edit, a diffusion-based TTS model capable of cloning anyone's voice with just a few seconds of audio. It comes in a 3.5B parameter flagship version and a 6GB lightweight variant that fits comfortably on consumer GPUs.
ByteDance just introduced DreamLite, a compact 0.39B diffusion model that handles both text-to-image generation and image editing directly on smartphones. It can generate a 1024×1024 image in around 3 seconds on an iPhone, swap styles, and remove objects completely offline. While fine details still lag behind massive server models, it brings fast, private image manipulation right to local devices.
Instacart just detailed its Physical AI push with Caper Carts, running NVIDIA Jetson chips directly on smart shopping carts. By processing basket contents, scales, and store locations in real time, the carts deliver timely nudges before checkout—driving double-digit basket size lifts for retailers.
Developer Yatharth Sharma just open-sourced LuxTTS, an ultra-fast voice cloning model that requires only a 3-second audio sample. Fitting in under 1GB of VRAM and outputting 48kHz audio, it reaches 150x real-time speed on GPUs and outpaces real-time even on basic CPUs. High-fidelity voice cloning can now run on almost any budget laptop.
Cohere just dropped Transcribe, its first open-source speech recognition model designed for note-taking and voice analysis. Weighing in at 2 billion parameters, it runs on consumer GPUs, processes 525 minutes of audio per minute, and supports 14 languages. It topped the Hugging Face Open ASR leaderboard with a 5.42 WER, and Cohere is also offering free API access.
Uber just announced a deal to invest up to $1.25 billion in Rivian, aiming to deploy 10,000 autonomous R2 vehicles by 2031. Starting in San Francisco and Miami in 2028, the robotaxi fleet will run exclusively on Uber's app. Why it matters: Uber wants to lock in hardware partners and cement itself as the platform for autonomous fleets before self-driving rivals push it aside.
Unsloth just launched Unsloth Studio, an open-source local UI that turns PDFs or CSVs into datasets and fine-tunes models 2x faster with 70% less VRAM. It packages visual dataset creation, training, and GGUF export for 500+ open models into one interface, and is available now on GitHub.
Memories.ai just partnered with Nvidia to build a dedicated visual memory layer for smart glasses and robots. Powered by Nvidia's Cosmos-Reason 2, the system indexes first-person video so physical devices can actually recall what they see, tackling a major blind spot left by text-only AI memory.
Nvidia just launched NemoClaw at GTC, a dedicated software stack for the OpenClaw agent platform. It packs Nemotron models and a new OpenShell isolated runtime into a single-command install, allowing autonomous agents to run 24/7 on local RTX PCs or DGX systems with strict privacy guardrails for cloud routing. Jensen Huang notably dubbed OpenClaw the OS for personal AI.
