Vannarot Roeung
Recent posts by Vannarot Roeung
Runway Bets on Ecosystem Over Product at AI Summit
One week after OpenAI killed Sora, Runway launched a $10M venture fund and Builders program at its first AI Summit. The company is building a platform for AI video, not just a tool.
Adobe Illustrator Turntable Turns 2D Vectors Into 3D
Adobe Illustrator Turntable is now generally available, automatically generating up to 74 editable 3D views from a single 2D vector.
Mistral Bets $830M on European AI Sovereignty
Mistral raises $830M in debt financing to build a Paris data center with 13,800 NVIDIA GB300 GPUs. Why Europe's biggest sovereign AI infrastructure play matters for creative professionals.
Qwen3.5-Omni Handles Text, Image, Audio, Video in One Model
Alibaba releases Qwen3.5-Omni, a multimodal AI model that processes text, images, audio, and video while generating real-time speech output in 36 languages.
Skywork Matrix-Game-3.0: Open-Source Real-Time Video AI
Skywork AI releases Matrix-Game-3.0, an open-source model that generates real-time interactive video at 720p and 40fps.
Claude Paid Subscriptions More Than Double in 2026
Anthropic confirmed that Claude paid subscriptions have more than doubled in 2026, with credit card data showing record subscriber growth in January and February.
Apple LGTM Brings 4K 3D Scene Rendering Without Per-Scene Training
Apple researchers published LGTM on March 28, 2026, a method that enables 4K-resolution 3D scene rendering using Gaussian Splatting without requiring per-scene optimization.
NVIDIA Kimodo: Open-Source Motion AI for 3D
NVIDIA has open-sourced Kimodo, the largest controllable motion diffusion model ever trained. Trained on over 700 hours of professional motion capture data, it generates 3D human and robot motions from text prompts.
The AI Music War: How One Week Fractured an Industry
In one week, Suno v5.5 added voice cloning, Udio pivoted to a walled garden, Google launched Lyria 3 Pro for developers, and Spotify built artist protection tools, fracturing AI music into competing visions.
Meta SAM 3.1 Doubles Video Tracking Speed for Creators
Meta releases SAM 3.1 with object multiplexing that tracks 16 objects simultaneously at 32fps on a single H100 GPU.
Deezer Licenses AI Music Detection to Rights Body
Deezer has licensed its AI-generated music detection technology to Hungarys EJI performers rights body. The platform flags over 60,000 AI tracks daily, with 39% of all uploads now involving AI.
Topaz Starlight 2.5 Adds Pro Video Upscaling to ComfyUI
Topaz Starlight Precise 2.5 is now available as ComfyUI partner nodes, delivering sharper video upscaling from 720p to 4K with fewer artifacts.
Claude Mythos Leak Signals a Step Change for AI
A misconfigured data cache exposed Anthropic's biggest secret: Claude Mythos, an AI model the company says represents a step change over Claude Opus 4.6 in coding, reasoning, and cybersecurity.
Anthropic Claude Mythos Leaks as Next-Gen AI Model
Anthropic accidentally exposed Claude Mythos, described as its most powerful model ever, through an unsecured data cache. The model shows major advances in coding, reasoning, and cybersecurity.
Shopify Tinker Puts 100 AI Tools in One Free App
Shopify launched Tinker, a free mobile app bundling 100+ AI tools for product photos, videos, logos, and 3D content. No prompt expertise required.
Suno v5.5 Adds Voice Cloning and Custom AI Music Models
Suno v5.5 brings voice cloning, up to 3 custom models, and My Taste personalization to its AI music platform.
Seedance 2.0 Relaunches on CapCut With IP Guards
ByteDance is rolling out Dreamina Seedance 2.0 globally through CapCut after pausing the launch to address copyright concerns from Hollywood studios. The model now blocks real-face video generation.
Voxtral TTS: Open Weights Challenge ElevenLabs
Mistral Voxtral TTS is a 4B parameter open-weights model that matches ElevenLabs quality in human evaluations, with 3-second voice cloning across 9 languages and self-hosting support.
Notion 3.4 Adds Custom AI Skills and Image Gen
Notion 3.4 introduces Custom Skills that turn repetitive AI tasks into reusable commands, AI image generation directly inside documents, and a new Presentation mode.
Windsurf Wave 13 Ships Free SWE-1.5 and Arena Mode
Windsurf released Wave 13 on March 6, 2026, delivering its SWE-1.5 coding model for free to all users, a new Arena Mode for blind model comparison, Plan Mode for structured task planning, and parallel multi-agent sessions.
Gemini 3.1 Flash Live Makes Voice AI Natural
Google launched Gemini 3.1 Flash Live, a new audio-first AI model designed for real-time voice conversations. It supports over 90 languages, filters background noise more effectively, and maintains context for twice as long.
Cohere Transcribe Tops Open ASR Leaderboard
Cohere released Transcribe, a 2-billion-parameter open-source speech recognition model that tops the HuggingFace Open ASR Leaderboard with a 5.42% average word error rate, beating OpenAI Whisper Large v3 by 27%.
Intel Arc Pro B70 Ships 32GB GPU at $949 for Local AI
Intel launches the Arc Pro B70 with 32GB GDDR6 VRAM and 367 INT8 TOPS at just $949, giving AI creators an affordable alternative to NVIDIA RTX Pro workstation GPUs.
ComfyUI Dynamic VRAM Ends Memory Crashes for Local AI
ComfyUI introduces a custom PyTorch memory allocator called Dynamic VRAM that eliminates out-of-memory crashes and lets creators run large AI models on limited hardware.