Toonflow is an open-source, multi-agent pipeline that turns raw novel text into animated short dramas — scriptwriting, storyboarding, and video generation included
Grok 4.5 tops Artificial Analysis's AutomationBench-AA with a 51% score, completing more SaaS workflow objectives than any rival at one-quarter the cost per task.
Apple open-sources AXLearn, its production JAX training system that runs the same code across GPUs, TPUs, and AWS Trainium with near-zero refactoring cost
Robbyant's LingBot-World is a fully open-source interactive world simulator that generates controllable video environments in real time, with sub-second latency and up to 10 minutes of coherent generation.
DeepTeam is an open-source Python framework that automates adversarial attacks on LLMs and AI agents, covering 50+ vulnerabilities and aligning with OWASP, NIST, and MITRE standards out of the box.
Pika's Director's Suite is an invite-only, Claude-powered agent that takes a video from concept to final cut — handling storyboards, character refs, clip generation, and timeline editing through chat or voice.
Krea adds Seedream 5.0 Pro to its platform, bringing pixel-level editing, multi-layer separation, and 10-reference image fusion to designers and developers
Artificial Analysis launches a controlled voice cloning leaderboard that finally separates model quality from voice selection bias, with Cartesia Sonic 3.5 taking the top spot
Cognition's SWE-1.7 matches frontier models at $1.97 per task, running at 1000 tokens/second inside Devin — powered by a novel multi-continent RL training pipeline.
Software Mansion's react-native-executorch lets you drop LLMs, vision, and speech models into any React Native app with a few hooks — no cloud, no native expertise needed.