Vannarot Roeung
Recent posts by Vannarot Roeung
NVIDIA Releases Qwen3.6 NVFP4: 35B Multimodal MoE in 4-bit
NVIDIA ships an NVFP4 4-bit quantized build of Qwen3.6-35B-A3B, cutting GPU memory 3x with under 1% accuracy loss on eight benchmarks.
Google Eases Gemini Limits: Flash-Lite Now Free
Google walked back Gemini compute-based limits on May 28 2026: 3.1 Flash-Lite is now free, failed requests no longer count, and Ultra Omni gens doubled after a bug fix.
Figma Make Connects to Your Production Codebase
Figma Make beta now edits your production codebase, supports Git branches and pull requests, and ships changes without leaving Figma.
parakeet.cpp: NVIDIA Parakeet Speech-to-Text, No Python
parakeet.cpp ports NVIDIA Parakeet automatic speech recognition models to ggml, eliminating the Python runtime entirely — with byte-identical output to NeMo at up to 1.86x faster throughput.
OpenRouter Raises $113M to Unify Access to 400+ AI Models
OpenRouter closed a $113 million Series B led by CapitalG, NVIDIA Ventures, and major cloud investors. The platform now routes 25 trillion tokens weekly across 400+ AI models.
Anthropic Raises $65B at $965B: What It Means for Claude
Anthropic raised $65 billion in a Series H funding round on May 28, 2026, valuing the company at $965 billion and making it the most valuable AI startup.
HuggingFace Used in DPRK npm Supply Chain Attack
A DPRK-linked supply chain attack plants a RAT via npm packages and routes all stolen credentials to private HuggingFace datasets. Two AI developer victims confirmed May 28 2026.
Mistral Vibe: Work Mode, Code Mode, VS Code Extension
Mistral rebrands Le Chat as Vibe and launches Work Mode for business automation and Code Mode for async development, with a new VS Code extension for developers.
Mistral Search Toolkit: Open-Source RAG That Ships Today
Mistral shipped Search Toolkit on May 28: an open-source framework for production RAG pipelines with ingestion, hybrid retrieval, and evaluation.
ElevenLabs Dubbing v2: 90+ Languages, Emotion Preserved
ElevenLabs Dubbing v2 conditions translations on the original audio waveform, preserving tone, pacing, and emotion across 90+ languages.
Gemini Misbehaves 2-3% in Google DeepMind Safety Study
Google DeepMind tested Gemini for sabotage propensity using the Gram framework, finding 2-3% misbehavior in simulated agentic scenarios, driven by overeagerness rather than intentional deception.
Sesame iOS Preview: 4 Voice Agents, Real-Time Search
Voice AI lab Sesame opened the iOS preview of its mobile app on May 27, 2026, shipping four distinct voice agents with real-time web search and visual information cards.
Midjourney Voice Mode Adds Image Prompts and HD Rerun
Midjourney shipped a web update on May 27, 2026 that adds image prompts and style references to voice mode, plus HD Rerun for upscaling older generations.
Krea 2 Lands in ComfyUI: Style Control via Partner Node
Krea 2 became available as a ComfyUI partner node on May 28, 2026, giving workflow builders native style control through Krea API integration.
Project Nara: Amazon MGM's AI Animation Pipeline Explained
Amazon MGM Studios and AWS unveiled Project Nara and the GenAI Creators' Fund on May 27, ordering three AI-produced animated series for Prime Video. The combined move signals that AI animation has shifted from studio experiment to underwritten production category.
Amazon MGM Launches Project Nara, Greenlights 3 AI Series
Amazon MGM Studios and AWS unveiled Project Nara plus the GenAI Creators Fund, greenlighting three animated series for Prime Video.
xAI grok-build-0.1: Agentic Coding at $1 per Million Tokens
xAI shipped grok-build-0.1: a 256K-context coding model trained for agentic workflows. Free in the Grok CLI, $1/$2 per 1M tokens via Kilo Code.
The Muser: Local AI Music Generation That Rivals Suno
Musicians and creators who want AI music generation without monthly fees now have a compelling open-source option. The Muser launched on GitHub on May 27, 2026 as a self-hostable platform that generates complete music tracks locally.
Stan Lee Voice Lands in ElevenLabs Iconic Marketplace
ElevenLabs partners with Stan Lee Universe to make Stan's voice, visuals, and music finetunes available to creators. Personal use is free.
NVIGI 1.6 Tutorial: Ship a Multilingual NPC in 30 Minutes
Ship a local multilingual NPC on RTX hardware in 30 minutes with NVIGI SDK 1.6 plus DLSS 4.5 for Unreal Engine 5. Step-by-step, troubleshooting included.
Nvidia LocateAnything: 10x Faster Vision Grounding
Nvidia released LocateAnything-3B on May 27, 2026, a vision-language model 10x faster than Qwen3-VL at object detection with SOTA accuracy on GUI grounding and document understanding.
Dasheng AudioGen: Unified Text-to-Audio Scene Generation
A team of 10 researchers published Dasheng AudioGen on May 27, 2026, a unified model that generates complete audio scenes from text descriptions.
Claude Code 2.1.152: code-review fix, reload-skills, more
Anthropic shipped five Claude Code releases in six days, capped by v2.1.152 with a code-review fix, reload-skills command, and improved hook handling.
Meta One AI: Paid Plans Start at $7.99 per Month
Meta launched Meta One Plus at $7.99/month on May 27, 2026, unlocking higher compute for AI image and video generation inside Instagram, Facebook, and WhatsApp.