Vannarot Roeung
Recent posts by Vannarot Roeung
HappyHorse: Alibaba's Stealth Video AI Tops Every Benchmark
HappyHorse-1.0, the anonymous AI video model that topped every benchmark, is from Alibaba. Analysis of Zhang Di's defection from Kling, single-pass architecture, and open-source claims.
MiniMax CLI Brings AI Video, Music, and Image Generation to Terminal
MiniMax released an official open-source CLI tool that brings AI video, music, image, and speech generation to the terminal with seven core capabilities.
llama.cpp Ships Multi-GPU Tensor Parallelism
llama.cpp b8738 adds backend-agnostic tensor parallelism that enables large AI models to run across multiple GPUs without vendor-specific code.
Black Forest Labs: The 70-Person Giant Killer Pivots
Black Forest Labs declined xAI partnership and pivots to physical AI. Analysis of the 70-person startup behind FLUX, its $300M+ enterprise deals, and what robots mean for creators.
GaussiAnimate Auto-Rigs 3D Gaussian Splatting
Researchers released GaussiAnimate, a framework that automatically rigs and animates 3D Gaussian Splatting assets with 17.3% quality improvement.
Luma Dream Brief Tests AI Ads at Cannes Lions
21 AI-generated ads compete at Cannes Lions 2026 with a $1M prize. Analysis of Luma's Dream Brief, director-AI collaboration, and post-DM9 transparency.
MegaStyle Trains FLUX on 1.4M Styled Images
Researchers from Tongji University, Tencent, and five other institutions released MegaStyle, a 1.4-million image dataset for style transfer alongside a FLUX-based model.
Waypoint-1.5 Generates Interactive AI Worlds on Consumer GPUs
Overworld AI released Waypoint-1.5, a real-time video world model that generates interactive environments at 720p and 60 FPS on consumer GPUs like the RTX 3090.
PrismAudio Turns Silent AI Video Into Stereo
PrismAudio, developed by Alibaba FunAudioLLM team, generates spatial stereo audio directly from silent AI video files in an average of 0.63 seconds.
LPM 1.0 Generates Real-Time Character Video
A team of 23 researchers released LPM 1.0, a 17-billion parameter Diffusion Transformer that generates real-time character video from audio input.
Adobe Firefly Adds Precision Flow and AI Markup
Adobe has launched two new AI image editing features in Firefly: Precision Flow generates multiple variations from a single prompt, while AI Markup lets creators draw on images to guide AI edits.
Suno Licensing Talks With Universal and Sony Stall Over Download Rights
Suno licensing negotiations with Universal Music Group and Sony Music have stalled over a core disagreement: whether users can download and share AI-generated songs outside the platform.
ChatGPT Pro Adds $100 Tier With 5x Codex Limits
OpenAI launched a new $100/month ChatGPT Pro tier on April 9, delivering 5x more Codex usage than Plus and filling the gap between the $20 and $200 plans.
ElevenLabs Brings Voice AI On-Premise and On-Device
ElevenLabs has announced on-premise and on-device deployment options for its voice AI platform, letting organizations run text-to-speech inference entirely within their own infrastructure.
Figma Weave: Node Graphs Replace the Prompt Box
Figma Weave connects 20+ AI models from 12 providers into a node-based canvas. Analysis of multi-model orchestration, App Mode, and what it means for design teams.
Meta Muse Spark Marks Proprietary Shift for AI
Meta launched Muse Spark on April 8, the first AI model built by its new Superintelligence Labs division. Unlike every Llama release before it, Muse Spark is proprietary.
PixVerse C1 Brings Cinematic AI Video Generation
PixVerse released C1, a new AI video generation model focused on cinematic production quality, on April 7. The model adds motion control, style transfer, and cinematic presets.
Anthropic Launches Claude Managed Agents for Production AI Workflows
Anthropic launched Claude Managed Agents in public beta, a cloud-hosted infrastructure service providing sandboxed execution, session persistence, and MCP tool integration at $0.08 per session-hour.
Cursor Bugbot Hits 78% Resolution With Learned Rules
Cursor's AI code review tool Bugbot now resolves 78% of the bugs it flags, a 26-point jump from its 52% rate at launch in July 2025.
Canva Acquires Simtheory and Ortto for AI Workflows
Canva has acquired Simtheory, an AI workspace for building custom agents, and Ortto, a marketing automation platform with over 11,000 customers across 190 countries.
Genspark Workspace 4.0 Adds Desktop Computer Use and Office Plugins
Genspark launched AI Workspace 4.0 with Claw for Desktop, bringing full computer use capabilities and native Microsoft Office plugin integration.
LM Studio Acquires Locally AI for Mobile
LM Studio acquired Locally AI, a mobile app for running large language models on iPhone, iPad, and Mac, expanding its local AI platform across devices.
X Launches Grok-Powered Photo Editor for Posts
X has rolled out a Grok-powered photo editor that lets users modify images through natural language prompts directly in the post composer. Currently available on iOS for Premium subscribers.
HeyGen Avatar V: AI Avatars Cross the Uncanny Valley
HeyGen Avatar V scores 0.840 Face Similarity from a 15-second webcam clip, beating Google Veo 3.1 by 17.6%. Full analysis of the video-reference architecture and creator impact.