01 Editors pick — latest

Gradium AI Releases New Default TTS Model: 81.0% Hard-Case Pass Rate at 216 ms Time-to-First-Audio

Speed and accuracy usually pull against each other in text-to-speech. Gradium AI's new default model reports both: an 81.0% human-rated pass rate on 500 hard sentences across five languages, at 216 ms P50 time-to-first-audio on Coval.

Michal Sutter  ·  Aug 31, 2026  Read the story » 02 Editors pick — story 02

Keenable AI Open-Sources NEEDLE: A Live Search Benchmark That Rebuilds Its Query Set Every Hour

How do you benchmark a web search API when the thing being tested can read the answer key? NEEDLE regenerates its query set every hour.

Asif Razzaq  ·  Aug 31, 2026  Read the story » 03 Editors pick — story 03

Google AI Releases TimesFM-3: A 330M Parameter Zero-Shot Foundation Model For Multivariate Time Series Forecasting

A 330 million parameter model that forecasts multiple related series in a single forward pass. The weights ship under a non-commercial licence.

Michal Sutter  ·  Aug 31, 2026  Read the story » 04 Editors pick — story 04

OpenClaw Releases OpenClaw 2.0: Guided Model Setup, 575 ms Control UI Startup, and One Trust Boundary Per Gateway

933 contributors and more than 16,000 pull requests. The rebuilt Control UI cut startup from about 1.6 seconds to 575 ms.

Michal Sutter  ·  Aug 30, 2026  Read the story » 05 Editors pick — story 05

Lowest-Latency Inference APIs for Voice and Realtime Agents: A Time to First Token TTFT-First Benchmark

Voice agents fail on latency long before they fail on intelligence. Every layer of the stack, with each figure labelled by how it was measured.

Asif Razzaq  ·  Aug 30, 2026  Read the story »
> /scale_b2b_pipeline
1M+
Active AI Builders

The high-ROI channel engineered to turn developer attention into validated enterprise pipeline.

Top-Tier Growth Teams Scale Here
NVIDIA•DeepMind•Meta AI•Anthropic
Reach The Devs
Promote GitHub
Launch Webinars
Feature Papers
Drive Pipeline
Weights & code

Open Source

 Licence noted on every entry
Latest 15
Keenable AI

Keenable AI Open-Sources NEEDLE: A Live Search Benchmark That Rebuilds Its Query Set Every Hour

How do you benchmark a web search API when the thing being tested can read the answer key? NEEDLE regenerates its query set every hour, which turns contamination into a scheduling problem rather than a disclosure one.

Aug 31, 2026Open Source
02 Google AI Releases TimesFM-3: A 330M Parameter Zero-Shot Foundation Model For Multivariate Time Series ForecastingGoogleNon-commercial 03 OpenClaw Releases OpenClaw 2.0: Guided Model Setup, 575 ms Control UI Startup, and One Trust Boundary Per GatewayOpenClawOpen source 04 Google AI Introduces EnvHarness: A Programmable Layer That Turns Static Agent Environments Into Adaptive Training WorldsGoogle CloudApache-2.0 05 Vercel AI Open-Sources vgpu: A TypeScript WebGPU Library for AI Agent ShadersVercelOpen source 06 Liquid AI Open-Sources Pipette: A Reproducible Benchmarking Suite That Measures On-Device Models, Quantization, Runtime and Hardware TogetherLiquid AIOpen source 07 Liquid AI Releases LFM2.5-DSpark Draft Models That Deliver Up to 3.18x Faster Decoding Without Changing Model OutputsLiquid AIOpen weights 08 NVIDIA Releases TensorRT Model Connect in Public Preview: Hugging Face Checkpoint to Native C++ Inference in Two CommandsNVIDIAApache-2.0 09 Meet SAM (Sovereign Agent Mesh): A Zero-Config, Zero-Trust P2P Network for AI AgentsGoogleApache-2.0 10 Nous Research Ships Bot Mode for Hermes Agent, Turning Agent Profiles Into a Roster of Named BotsNous ResearchMIT 11 MiniMax Releases MiniMax-Music3: An Open-Weights Music Model Generating Complete Five-Minute Songs From Lyrics and a Structured CaptionMiniMaxOpen weights 12 DeepSeek AI Releases DeepSeek Harness in Developer Preview: An MIT-Licensed Agent Harness Where Everything is a PluginDeepSeek AIMIT 13 Liquid AI Releases LFM2.5-VL-3B: A 3B Vision-Language Model That Reads Screens, Grounds Objects, and Calls Tools On-DeviceLiquid AIOpen weights 14 NVIDIA AI Releases Nemotron 3.5 Lightning: A 30B Open MoE with 3B Active Parameters, and NeMo Switchyard Model RouterNVIDIAOpen weights 15 Meta AI Releases Muse Glimmer: A 30B Open-Weights Agentic Model That Runs on One Consumer GPUMeta AIApache-2.0
Browse the full Open Source archive →
Robots & VLA

Physical AI

 Latest 10 · spec on every entry
Robbyant / Ant Group

Ant Group's Robbyant Unveils LingBot-VA 2.0: A Causal Video-Action Model Built Natively for Physical AI

Built from scratch for embodiment rather than fine-tuned from a video generator. It predicts the next state before executing, then re-grounds on what it actually sees.

225 Hzasynchronous control ASIF RAZZAQ   ·   JUL 11, 2026
MirrosMeet ‘Code-as-World’: An Agentic Loop That Rewrites Real Videos Into Executable MuJoCo Physics Programs MuJoCo Hugging FaceHugging Face Unveils Microduck: A $399 Open-Source 25 cm Biped You Train with Reinforcement Learning $399 Dyna RoboticsDyna Robotics Introduces Dyna-2: A World-Action Model Pre-Trained on 1 Million Hours of Human Video 1M hours NVIDIANVIDIA Releases Alpamayo 2 Super: A 34B Open Vision-Language-Action Model for Robotaxis and Autonomous Driving 34B Google DeepMindGoogle DeepMind Ships Three Physical AI Models For Whole Body Control, Dexterity And Multi Robot Collaboration 3 models NVIDIANVIDIA Releases Cosmos 3 Edge: A 4B-Parameter Open World Model That Reasons and Generates Robot Actions On-Device 4B edge NVIDIANVIDIA Released DeepStream 9.1: Bringing Agentic AI to Vision AI With 13 Skills and Multi-View 3D Tracking 13 skills Mistral AIMistral AI Releases Robostral Navigate: An 8B Model Enabling Robots to Navigate Complex Environments Using a Single RGB Camera 8B RobbyantRobbyant Releases LingBot-VLA 2.0: An Open-Source 6B Vision-Language-Action Model for Cross-Embodiment Robot Manipulation 6B
Browse the full Physical AI archive →

Content from Our Partners

Content from Our Partners

PROMOTED LATEST 15 PLACEMENTS
01 PRINCETON / ANT GROUP / STANFORD Researchers from Princeton, Ant Group and Stanford Introduce AQuA: A Two-Part Agentic Framework for Autonomous Factor Discovery and Model Development in Quantitative Finance SEP 01, 2026 02 UPDF Meet UPDF: A Lightweight Adobe Alternative Built for the Agentic Era AUG 20, 2026 03 NVIDIA The Video Production Stack Now Fits on One Desk: LTX-2.5 Launches as NVIDIA-Accelerated Open Weights World Model AUG 11, 2026 04 COPILOTKIT CopilotKit Open Sources Channels SDK: An MIT Licensed Library That Runs Any AG-UI Agent Inside Slack And Microsoft Teams AUG 04, 2026 05 MEND.IO How to Secure AI Agents, MCP Servers, and LLM Apps in Production AUG 03, 2026 06 DATALAB Datalab Marker v2 vs MinerU, Docling, and Liteparse: Benchmark Breakdown JUL 24, 2026 07 MONGODB Build an Agentic Event Venue Operator with MongoDB Atlas, Voyage, and LangGraph JUL 17, 2026 08 ROBBYANT Ant Group's Robbyant Unveils LingBot-VA 2.0: A Causal Video-Action Model Built Natively for Physical AI JUL 11, 2026 09 ROBBYANT Meet LingBot-World-Infinity: An Open Causal World Model With An Agentic Harness JUL 09, 2026 10 ROBBYANT Robbyant Releases LingBot-VLA 2.0: An Open-Source 6B Vision-Language-Action (VLA) Model for Cross-Embodiment Robot Manipulation JUL 08, 2026 11 ROBBYANT Ant Group's Robbyant Open-Sources LingBot-Vision: A 1B Boundary-Centric Vision Foundation Model for Dense Spatial Perception JUL 07, 2026 12 WEBBRAIN Meet WebBrain: An Open-Source, Local-First AI Browser Agent That Reads Pages and Automates Tasks in Chrome and Firefox JUL 02, 2026 13 DATALAB Datalab Releases lift: A 9B Open-Weights Vision Model That Extracts Structured JSON From PDFs Using Schemas JUN 23, 2026 14 ATOMS Meet Atoms: A Vibe Coding Tool That Uses AI Agents to Build, Deploy, and Market Your App (No Code) JUN 16, 2026 15 TINYFISH TinyFish Launches BigSet: An Open-Source Multi-Agent System That Builds Structured Live Datasets from Plain-English Descriptions JUN 02, 2026