Toravo AI News · Toravo Lab
Tools we featured
Every tool from our AI News Shorts and Lab tests, with the official links. You commented TOOL? Here you go. Newest first. Tested, not hyped.
Kimi K3.1 Pro Weights Are Out – Open, But Huge
Hacker News front page · huggingface.co
Suno Now Starts Songs From MIDI – Available for Pro and Premier
Qwen 3.8 Reads Raw Model Weights on a Laptop – After 46 Minutes
ComfyUI Gets Idol: Open Image-to-3D Model, Apache-2.0
ChatGPT Share Link Could Take Over Your Account – Now Fixed
Claude Ultra: No Five-Hour Limit for Verified Accounts
ElevenLabs Cue Aligns an Hour of Audio in Under a Second
Mistral Large 4 Is Open Weight – But Built for Coding
Pika Opens Video-and-Audio Model – But It's 210 GB
Genie 4 Builds Two-Minute Worlds – But You Can't Try It
MAI Image 2.7 Is No. 2 for Image Editing, Behind GPT Image 2.5
Microsoft AI · artificialanalysis.ai
SpAItial Echo-3 Turns Your iPhone Scan Into a 3D Room
Nvidia Releases Free Transcription Model – No Live Audio
Qwen Model Tried to Call Home to Beijing – Cohere Found It
Genie 4 Adds Audio – But Consumer Access Starts at $100 a Month
OpenAI Cut One Model's Price by 80% – Why?
Phonon 31M English: Apache 2.0 Voice Model Runs on a Laptop CPU
Kimi K3.1 Pro: Open Weights Promised, Not Here Yet
Hy-Nomas Gives Open Video Camera Control, Only 49 Downloads
Block's Dessa Voice Model Runs Locally – Not Public Yet
Dessa Listens and Talks at Once – But Not for You Yet
Kling 4.0 Makes 30-Second Video Clips – Here's the Limit
HeyGen Video 1 Makes Full AI Clips With Sound – Here's the Catch
Ming TTS Lands in ComfyUI – With Three Likes So Far
Canva Launches Its Own Image Model – Cheaper Versions Next
Adobe Lightroom Opens Closed Eyes in One Tap, But Free Gets Two Tries
Flint Releases Open Coding Model – But Needs a B300 or H200
Ideogram Turns Images Into Editable SVG Art
ideogram.ai · developer.ideogram.ai
FastH3 Omni: 8-Step Video Model With Sound for ComfyUI
Spotify Removed 50 Million AI Slop Tracks From Recommendations
PHP Flaws Found by AI: Sites Go Dark Until Patched
Fibo 2.0 Lands in ComfyUI – But Only for Non-Commercial Use
FLUX 3 Max: Seedance-Level Video in About 12 Seconds
Nemotron Dory 4B: vLLM Support Isn't Released Yet
DeepSeek Drops Open Vision Model – Skip It for Music
Ling-3.1-flash Is Open – But It Needs Eight H200 GPUs
Open Jev Model Hit 171,832 Downloads – Not the Closed One
ElevenLabs Swaps the Main Character – But Only on Paid Plans
ElevenCreative · elevenlabs.io
MOSS-Image in ComfyUI – Repackaged Model Files
FLUX 3 Image Klein Is Open – But One User Reports Distorted Eyes
ComfyUI Adds Nvidia Upscaling, Frame Generation and HDR Nodes
Fish Audio Drama 3 Shrinks – But It's an Unofficial Quant
Grok and Cursor Now Share One Usage Pool – Here's the Catch
REA: An AI Agent That Reverse-Engineers Apps
EmbeddingGemma 2 Transcribes Speech – But 5% Error Rate
Can an Open Lip-Sync Model Run on One RTX 4090? LongCat-Video-Avatar
LongCat-Video-Avatar (Hugging Face) · LongCat-Video-Avatar-1.5 (Hugging Face) · LongCat-Video-Avatar 1.5 Technical Report · LongCat-Video (base model) · GitHub meituan-longcat/LongCat-Video (README, issues #19, #52, #79)
PDF TO MP3 on Show HN: Use Your OpenAI Subscription for Audiobooks
Qwen-Image 'Uncensored' Lands — No Uncensoring Effect On Images
ElevenLabs Saves Your Brand Style – You Still Tag It
DistroKid Is Pulling Your Songs With No Warning
Qwen3-1.7B Routes in One Pass – But It Lies
Gemini Nano Banana 2.1 Just Hit ComfyUI
Does the Expensive Pipeline Fix AI Video Flicker? We Measured It.
Lightricks/LTX-2 issue #223 – Artifacts in high-frequency regions when generating videos · LongCat-Video-Avatar (Hugging Face, next Lab)
Gemini 4 Argon Is Here – But Cyber Defenders Get It First
GPT-6.1 Sol at One-Fifth the Price – Its Big Sibling Reportedly Axed
Suno Just Joined the TTS Race — With Music Bundled In
ElevenLabs Doubles to $22B – and Employees Cash Out $300M
OpenAI Unveils a Jev Look-Alike – But Only as a Preview
Comfy API: Your ComfyUI Workflow as a Product Endpoint
MiniMax H3 Character Swap LoRA - First Look for Creators
Laya Multilingual: 322M Decision Model for 100+ Languages
Suno download cap: what creators need to know
YuE2 Skill v1.2: Pure Instrumentals From Text
Laya — Tiny Open-Source Decision Model for AI Pipelines
Everyone Hypes Fast AI Decision Models. We Tested One.
TypeSafe AI – Introducing System One Models and Jev · Laya (Hugging Face)