explainx / blog / topics
AI Image, Video and Voice
Generative media models make images, video, music, and voices. New releases arrive constantly, along with questions about rights, detection, and quality.
This page tracks the image, video, and voice models and tools we covered.
138 stories · latest Nov 18, 2026
Start here
Multimodal Vision Token Budget: How Images Consume Context Windows
A 128K context window is not 128K words when you attach screenshots. Vision encoders chop images into tiles that bill like text tokens, often hundreds per frame. This guide explains the math, provider patterns, and practical trims for agents that send UI captures or camera frames.
One Model per Camera: AWS Shows a SageMaker SDK v3 Pipeline for Edge Vision Fleets
A fleet of factory cameras rarely shares one model. An AWS Machine Learning blog post from October 26, 2026 shows a SageMaker SDK v3 reference pipeline that trains an anomaly detector and a classifier per camera, gates them on metrics suited to imbalanced defect data, compiles for the edge with Neo, and paces thousands of retrains.
The Storm Dance Trend: What It Is and How to Make Your Own AI Version
A rapid-fire arm-sweep-and-spin sequence timed to a bass drop is taking over short-form feeds under the name "the Storm." explainx.ai breaks down what actually defines the format, why it's spreading this fast, and gives a full workflow for filming your own or building an AI-enhanced version with Sora, Veo, Kling, and Runway.
How to make viral AI stunt videos like the Sylvester Stallone skate clips
Creator George Wu's hyper-realistic AI videos of Sylvester Stallone skating ramps at 80 blew up on X — while Jim Chuong's "Finally, something that isn't AI" quote-tweet hit 9.6M views as the punchline. explainx.ai walks through how these clips are made in PixVerse, how to label them honestly, and how viewers can tell real from generated without trusting vibes alone.
Build With MiniMax H3 Max: API, Local Setup, and 8 Project Ideas
H3 Max is fast enough to move AI video from a background render queue into an interactive product loop. This developer guide shows the hosted API path, explains what can and cannot run locally, and turns the speed gain into eight concrete applications.
Using Gemini to Spot Fake Cosmetics: A Multimodal LLM Case Study
Prof. William Grover photographed three packages of Rhode Peptide Lip Tint — two $5 eBay fakes and one genuine Sephora unit — and asked Google Gemini 3.6 Flash whether each was authentic. The model nailed both counterfeits by cross-referencing typos and mismatched compliance data across photos, then confidently declared the real one a fake. A clean look at where multimodal models help, where photo artifacts break them, and how to prompt around it.
Timeline
November 2026
Nov 18
Suno Visuals: H3 Move Music Videos, Lyric Clips, and Visual CreditsSuno’s November 17 Visuals release turns library songs into up-to-15-second clips, lyric videos, and shareable aspect ratios. Video uses H3 Move from MiniMax H3 with its own Visual credit pricing.
Nov 17
Bria RMBG-3.0: Open-Weights Background Removal Built on Fibo 2.0Bria released RMBG-3.0 as open weights on Hugging Face, distilling Fibo 2.0 into a four-step alpha-matte model. The model card claims top ranks on 234 hard cutouts versus Photoroom and remove.bg; commercial use needs a Bria license or API.
Nov 17
ElevenLabs Acquires Kotoba: Koto Japanese Voice Models Join ElevenAgentsElevenLabs announced on November 16, 2026 that it will acquire Kotoba Technologies, the Tokyo-rooted voice lab behind the Koto model family. The deal targets Japanese text-to-speech, speech-to-text, and simultaneous interpretation for ElevenAgents customers such as Salesforce Agentforce Voice.
Nov 15
First AI Music Awards in Los Angeles: What SIQA Built and Who WonSonic Intelligence Academy held the first AI Music Awards on November 13, 2026 at the Mayan Theater in downtown Los Angeles. Variety's on-site report describes a peer-voted ceremony, a verified AI disclosure registry, and a night of teleprompter and Deepgram voice glitches alongside real performances.
Nov 15
Sophon-Realtime v1: Open Full-Duplex Voice With a 32B Brain and vLLM-OmniSophon-Realtime v1 is a family of open full-duplex voice models from Adventists AI. A shared 32B brain handles listening, turn-taking, and replies; Tag, Flash, and E2E builds differ only in how speech is rendered. The model card claims VoiceBench 82.8, GPT-Live-compatible live mode, and pip-install serving on one 80 GB GPU.
Nov 14
Haru-4B: An Open, Apache-2.0 Full-Duplex Voice Model Built on Qwen3.5-4Bhigurashi-ai released Haru-4B, a pretrained full-duplex spoken dialogue model that listens and talks at the same time. It uses a Qwen3.5-4B backbone, continuous audio latents at 7.5 Hz and an Apache-2.0 license. explainx.ai covers how it works, how to run it and what the authors have not shown yet.
Nov 13
Open Vision Banana: An Open Recipe That Turns FLUX.2 Klein Into a Segmentation and Depth ModelA research team published Open Vision Banana, a fully open recipe that fine-tunes FLUX.2 Klein 4B and 9B image generators into one model for segmentation, depth and surface normals. It reports 73.9 mIoU zero-shot on Cityscapes against 69.9 for Google DeepMind Vision Banana.
Nov 13
PhysLoop: ByteDance and PKU Harness LLMs to Fix Physics in Video ModelsResearchers from ByteDance Seed, Peking University, UCSB, MIT, and partners published PhysLoop, a harness that guides large language models to generate and repair physically consistent videos inside simulators. The paper is out; the code and the 100K-clip dataset are still pending release.
Nov 13
Seedance 2.5 Flash: BytePlus Claims 2x Speed at $0.152 per Second. What Is Verified?On November 12, 2026, BytePlus announced Dreamina Seedance 2.5 Flash on X: results close to flagship Seedance 2.5, nearly twice the speed, and USD 0.152 per second at 720p. The API is due on ModelArk on November 15. Here is what is verified and what is still a vendor claim.
Nov 10
Fish Audio Live 3: Full-Duplex Voice Timing for Real CallsOn November 10, 2026 Fish Audio launched Live 3, a full-duplex voice model trained on real conversations so agents can acknowledge, wait, or yield mid-call. Here is what the company claims, how it differs from Drama 3 TTS, and how to try it.
Nov 10
MiniMax-H3-Image: Official 16 MP Open Weights and RepoMiniMax-AI released the official MiniMax-H3-Image repository and Hugging Face weights in November 2026: a 33.1B distilled DiT for 16 MP generation and multi-reference editing in six steps, with deployment guides for SGLang, vLLM, diffusers, and ComfyUI.
Nov 9
Multimodal Vision Token Budget: How Images Consume Context WindowsNov 8
Shizuku-Duplex: Japanese Full-Duplex Voice With 80 ms Ticks and Async ToolsOn November 8, 2026 Shizuku AI Labs published Shizuku-Duplex, a Japanese full-duplex dialogue model that ticks every 80 ms, keeps talking during backchannels, and hands web search to a second model without freezing the conversation.
Nov 8
Viggle Remake: Open 20B Video-to-Video Model With an Agent MCP ServerViggle AI released Viggle Remake on Hugging Face: two LoRA modes on the MiniMax-H3 base, a reference sampler, ComfyUI packs for smaller GPUs, and an MCP server that lets Claude Code or Codex cut windows, paint keyframes, and render a full soundtracked film. Here is how it compares to Animate, what the license requires, and which hardware fits.
Nov 4
Atmanity Atmee v2: A Live Talking Avatar From One Image, and the 50x Faster Claim CheckedAtmanity released Atmee v2, a real-time video model for live avatars, and Causeur, a behavior model. It claims big speed and cost gains over Wan 3.0 and Seedance 2.5. Access is by invitation, and the numbers are the company's own.
Nov 4
MoviEdit: Adobe and KAIST Edit Multi-Shot Video With One InstructionMoviEdit is an Adobe and KAIST research model that applies one edit consistently across every shot of a clip up to 25 seconds long. The team released the benchmark and reference results, but we found no model weights.
Nov 3
P-H3-Image: Pruna Turns MiniMax-H3 Into an Open 8-Step Image ModelPruna AI released P-H3-Image on Hugging Face: a 33.1B-parameter image generation and editing model derived from MiniMax-H3. It renders 8-megapixel images in 8 steps, edits with up to 8 reference images, and runs in 21 GB of VRAM in a quantized ComfyUI build. The license excludes the US, EU, UK and South Korea.
Nov 3
Salesforce to Buy Cresta: Voice Agents Come to Service CloudOn November 2, 2026 Salesforce said it signed a definitive agreement to acquire Cresta, a voice-first AI company for contact centers. The $2.7B figure comes from The Information, not Salesforce. We separate the confirmed facts from the reported ones and explain what it means for teams building customer-service agents.
Nov 2
Hume Audio Decision API: Ask Any Question About a Call, Get a Probability BackOn November 2, 2026 Hume launched the Audio Decision API: send raw audio, write a question and its possible answers, and get probabilities back. Hume says it agrees with a human panel 70.7% of the time. Here is what is documented, what is vendor-reported and what to check before you use it.
October 2026
Oct 30
Biological Computing Co.'s "Neuron-Derived" AI Video Model: 5x Faster, 80% Cheaper? Claimed vs VerifiedThe Biological Computing Co. and AWS pitch a "neuron-derived" text-to-video model said to be five times faster and 80% cheaper than its base model. explainx.ai separates the self-reported numbers from the one outside test and lists the questions still open.
Oct 30
WaiT: Meta Releases Code for a Wavelet-Aware Image Generator That Reports FID 1.43 on ImageNet 512 With Up to 50% Less Sampling ComputeA new facebookresearch repo accompanies the paper WaiT for the Signal. It denoises coarse structure first and lets fine detail join later, and the authors report a pixel-space FID of 1.43 on ImageNet 512 with up to 50% less sampling compute. We read the README and abstract.
Oct 30
Spotify AI Remix Tool for About $5 a Month in 2027? Claimed vs VerifiedA Financial Times report says Spotify wants to launch a paid AI product for remixes and covers in early 2027 at roughly $5 a month, pending licensing deals. We separate the report from what Spotify and Universal have actually announced.
Oct 29
Cartesia Managed Agents: #1 on Voice Agent Benchmarks at $0.03 a Minute, What the Numbers MeanCartesia launched Managed Agents, an end-to-end voice agent platform that says it ranks first on Artificial Analysis's Speech Agent Arena and ServiceNow's EVA-Bench, with 50 percent launch pricing of $0.03 a minute. We read the launch post, check what a third party confirms, and flag the caveats.
Oct 29
Deepgram Acquires Daily, Maker of Pipecat: What Changes for Voice Agent BuildersDeepgram announced on October 29, 2026 that it is acquiring Daily, the maker of the open-source Pipecat framework, and launching an open protocol called the Real-Time Context Standard. The price was not disclosed. Here is what is confirmed, what to watch, and how to protect your own stack.
Oct 29
Phonic-1: A New Speech-to-Speech Model Claims 600 ms and Tops One Index, Claimed vs VerifiedPhonic announced Phonic-1, a speech-to-speech model it calls the most advanced for agentic tasks, alongside a $27.5M Series A led by First Round. We checked the 600 ms response claim, the tool-calling claim and the pricing against Phonic's own pages and the Artificial Analysis speech-to-speech leaderboard.
Oct 29
Pika H3 Flex: An Open Video and Audio Model Built on MiniMax H3, in Six StepsPika released H3 Flex, two LoRA adapters that turn MiniMax H3 into a six-step video and stereo-audio generator. It ranks seventh on DesignArena text-to-video, runs in ComfyUI, and inherits MiniMax's community license. Here is what you get, what it costs, and the catches.
Oct 28
AssemblyAI-TTS-1: $15 per Million Characters and 90 ms to First AudioAssemblyAI, known for speech-to-text, has launched AssemblyAI-TTS-1 for voice agents. It prices at $15 per million characters with no concurrency tiers. Most numbers come from AssemblyAI itself; here is what is independently ranked and what you should test.
Oct 28
LTX-2 Marking Pack: Lightricks Adds Opt-In Invisible Watermarks to Open Video GenerationLightricks has published LTX-2-Marking on Hugging Face, a weights pack that lets the open LTX-2 pipelines embed an invisible PixelSeal mark in video and an AudioSeal mark in sound. It is opt-in, not cryptographic, and removable by a determined attacker. Here is how it works and what it means.
Oct 28
Rime Prism: Speech-to-Speech With a Text Checkpoint, and a 300ms Claim That Does Not Match Rime's Own 0.57sRime launched Prism on October 28, 2026: a speech-to-speech model that writes what it will say before it speaks, so guardrails can read and rewrite it. Secondary coverage quotes a 300ms response; Rime's own numbers are 0.57 seconds and about 500ms.
Oct 27
MAI-Image-2.7-Preview: Microsoft Claims No. 2 in Image Editing, What Is VerifiedMicrosoft AI previewed MAI-Image-2.7 on October 27, 2026, edit-only for now in the MAI Playground, with a reported No. 2 on Artificial Analysis image editing behind OpenAI. Here is the claim, the context from earlier MAI-Image releases, and the gaps.
Oct 26
One Model per Camera: AWS Shows a SageMaker SDK v3 Pipeline for Edge Vision FleetsOct 24
Kromoss: A Full Fine-Tune of MOSS-Image With 8-Step Turbo LoRAs, Apache 2.0Kromoss is a community full fine-tune of OpenMOSS's MOSS-Image-Base from lodestones, trained on 704k images and released under Apache 2.0 with two 8-step turbo LoRAs. It has been trending on Hugging Face. Here is how to run it and how far its one benchmark goes.
Oct 23
Block's Dessa: A Voice Model Trained on Speech Alone, With No Text, That Runs on a MacBookOn October 23, 2026 Block announced Dessa, a research-preview voice model it says was trained from scratch on nothing but human conversation audio, listens and talks at the same time, and runs locally on a MacBook Pro M4 Max. There are demos and a waitlist, but no paper, weights or benchmarks yet.
Oct 22
Canva Craft: First Image Model, Trained on TemplatesOn October 22, 2026 Canva released Craft, its first in-house image generation model. It plans the layout before it draws, is trained on human-designed templates rather than photos, and ships to every Canva user through Dream Lab and the Canva API. Here is what is confirmed, what is not, and how it fits Canva's wider AI push.
Oct 22
FastH3 Omni: Open 8-Step Reference-to-Video With Audio, From DGX Spark to B200The FastVideo project, with Nuva Lab and NVIDIA, released FastH3 Omni: an eight-step distilled version of MiniMax-H3 that takes a prompt plus up to twelve image, video and audio references and returns video with synchronized stereo audio. Weights and training code are open, under the MiniMax-H3 community license.
Oct 22
Gradium Phonon: A 31M-Parameter Open TTS That Runs in Your BrowserGradium released the Phonon family of tiny text-to-speech models. Phonon-31m is Apache 2.0, about 65 MB quantized, and streams speech from a browser tab or a laptop CPU. The larger 40m models are noncommercial.
Oct 20
Bria Fibo 2.0: Open-Weights 8B Image Model Ties for the Lead — $32 per 1K Images, Claimed vs VerifiedBria released Fibo 2.0, an 8B open-weights model that generates and edits images from structured JSON captions. Artificial Analysis lists it a hair behind Qwen-Image-2.1 for open text-to-image and first for open editing, but the weights are non-commercial and the $32 per 1,000 images figure is a hosted-API claim.
Oct 20
Eleven Realtime: ElevenLabs Ships Two Speech-to-Speech Models for Voice AgentsElevenLabs released Eleven Realtime on October 20, 2026: two end-to-end speech-to-speech models, v1 Mini for natural conversation and v1 Max for tool-calling tasks. Here is the benchmark table, how the OpenAI-compatible API works, which numbers are the vendor's own, and what to test before you switch.
Oct 20
LTX-2.5: Lightricks Ships a 22B Open-Weight Video and Audio Model With Native MultishotLightricks has put LTX-2.5 on Hugging Face: a 22B-parameter open-weight model that generates synchronized video and audio, now with native multishot generation. Here is what the model card says, what the license allows, and how to run it locally.
Oct 20
MT Forcing: Fixing Drift in Multi-Turn AI Image EditingAsk an AI image editor for five edits in a row and the picture usually decays: textures smear, faces change, earlier edits vanish. MT Forcing, a new paper with released code and checkpoints, trains the model on its own multi-turn rollouts to stop that drift. Here is what it does and what the numbers do and do not show.
Oct 20
OpenFilm: The Open-Source Video Agent That Turns Claude Code or Codex Into a Film StudioOct 20
RoAR-VAE: The Image Tokenizer Recipe Aiming at FLUX.2-VAEOct 20
Voxtral 3: Mistral Ships a Voice Agent SDK and New Speech ModelsOct 19
FLUX 3 Max on fal: Joint No. 1 Image-to-Video Claim, 75% Off Until November 7Oct 19
Kling 4.0 Launches: 30-Second Video, Stereo Audio and a Flash TierOct 19
Speculative Sampling Makes Big Video Models Interactive Without RetrainingOct 17
MOSS-Image: OpenMOSS Releases a 5B Open Text-to-Image Model With Layout ControlOct 16
Elorian Glimpse: A Visual Reasoning Model That Looks Back at the ImageOct 16
FLUX Video Upscale 4K Price Cut: $0.59 to $0.28 per Second, CheckedOct 16
Taco Bell Drive-Thru AI and Employee Retention: What Yum Brands Claims vs. What Is VerifiedOct 15
FLUX 3 Image Klein: Black Forest Labs Opens a 5.6B Image Model for 8GB GPUsOct 15
Krea 3: Image Editing in Under 8 Seconds at 2K for $0.06, Versus Nano Banana Pro and GPT Image 2Oct 14
Deezer Will Add a Toggle to Hide AI-Generated Music by the End of 2026Oct 14
Mirelo SFX 2.0 Replaces a Video's Sound but Keeps the Dialogue, Up to 10 MinutesOct 14
YouTube Music Adds "Made with AI" Labels to Songs: What Gets Tagged and What Does NotOct 13
Baroness Campbell Delivers the First AI-Voiced Speech in UK ParliamentOct 9
“Gods Don’t Give Gifts” Is the First AI Movie With an MPA RatingOct 9
Iris-3B: An Open Pixel-Space Image Model With No VAE, and an Honest Negative ResultOct 9
Voyager: The "Codex for Creative Work" Harness for 100+ Tools, ExplainedOct 6
Kandinsky 6.0 Video: A 29B Open Model That Makes Sound and Picture Together
September 2026
Sep 29
Kling 4.0 Preview Leak: Discord Screenshots, Not a LaunchSep 26
The Storm Dance Trend: What It Is and How to Make Your Own AI VersionSep 26
How to make viral AI stunt videos like the Sylvester Stallone skate clipsSep 24
Fish Audio Drama 3 Preview: Directing AI Voice in Plain Language Instead of Audio TagsSep 20
AI Poster Slop Isn't a Model Problem — It's a Prompting ProblemSep 17
ElevenLabs Reception: The Questions Small Business Owners Actually AskedSep 15
Brain Implant + AI Voice: What the Paralyzed-Speech Demo Actually ShowsSep 10
Suno Launches v6 AI Music Models With Warner and BMG After Licensing SettlementSep 10
This Tennis AI Coach Was Built With Roboflow Agent and Claude CodeSep 9
HyperFrames: HeyGen’s Open-Source HTML-to-Video Framework for AI AgentsSep 7
VoiceStudio: The Open-Source, Fully-Local ElevenLabs AlternativeSep 3
Ato Is a Screen-Free AI Device for Seniors. 2.9M Views and One Hard Question.Sep 2
Developer Builds Endless AI TV With MiniMax H3 MaxSep 2
Google Pics: Workspace's New AI Image Editor ExplainedSep 2
Build With MiniMax H3 Max: API, Local Setup, and 8 Project Ideas
August 2026
Aug 31
LeVJEPA: Video Pretraining at 20× Less Compute Than V-JEPA 2Aug 31
Motion.so: URL-to-Launch-Video Agent for Motion DesignAug 31
OpenShot 4.0: Local AI Masking, Color Grading, and Screen RecordingAug 29
Using Gemini to Spot Fake Cosmetics: A Multimodal LLM Case StudyAug 29
Diffusion Studio's open-source video editor turns every edit into codeAug 29
MiniMax Fast H3 v1: real-time open video on BlackwellAug 28
Digital Camouflage: The Shirt That Makes AI Cameras Blind to YouAug 28
fal's H3 Max generates video faster than you can watch itAug 28
PhoneLLM: An Open Voice-Agent Model Claiming GPT-5.6 Terra Quality at 1/18th the CostAug 27
World-First AI-Assisted Brain Surgery: What the AI Actually DidAug 26
Awesome GPT-Image-2: Prompt-as-Code Engine with 530+ Cases and Agent SkillsAug 25
UK Gets Ukraine Avengers AI Labs: 5M Battlefield Frames for Model TrainingAug 22
Nari Labs Hits Sub-50ms TTS at $2 per Million CharactersAug 19
BGRemover.video: AI Video Background Removal, No Green Screen NeededAug 18
Cartesia Sonic-3.6: #1 on Both Artificial Analysis TTS BoardsAug 18
Roboflow Benchmark: GPT-5.6 Sol Is OpenAI's Best Vision Model — Gemini Still WinsAug 18
Stable Audio 3.0 Gets a DAW Plugin and a Rebuilt Web AppAug 9
Higgsfield Offers 33 Days of Unlimited Seedance 2.5 VideoAug 5
Bland Speech v3: Inside the "Human Speech Engine" LaunchAug 1
RF-DETR: Roboflow's Real-Time Detection Transformer, Explained
July 2026
Jul 31
Hugging Face Speech-to-Speech: Build Open-Source Voice AgentsJul 31
MiniMax H3: Open Video Model — Locked Out of the US and EUJul 30
Google Earth + Nano Banana 2: Reimagine Any PlaceJul 29
Fish Audio Raises $52M and Launches S2.1 Pro Voice AIJul 26
Inflect-Micro-v2: Full Local TTS Under 10M ParamsJul 23
FLUX 3: Black Forest Labs Unifies Video, Audio, and Robot ActionJul 19
LingBot-Map: Streaming 3D Reconstruction at 20 FPS — Robbyant GCT Guide (2026)Jul 19
Skyroot Vikram-1 Mission Aagaman: How AI Was Used — Onboard, in Engineering, and to Understand the LaunchJul 15
Overtone: Hinge Founder's $18M AI Matchmaker With No Profiles or SwipesJul 10
Reve 2.1: #2 Text-to-Image Arena, Top 4K Model, and Layout-First Visual IntelligenceJul 9
Google Photos Video Remix: Gemini Omni AI Video Editing in the Create TabJul 8
Silent Speech with Ultrasound: Aleph Neuro's 15.6% WER Demo ExplainedJul 8
Kokoro TTS: Local CPU-Friendly Speech at 82M Parameters (HN Guide, July 2026)Jul 7
X iOS Video Editor: Overlay Captions, Green Screen, and In-App Recording (July 2026)Jul 4
Seedance 2.0 Korean Neighborhood Prompt: The 12M-View Recipe ExplainedJul 3
Can Claude or LLMs Watch a Video? Here's How to Make It Work
June 2026
Jun 27
AI for Creative Hobbies: Music, Art, Writing, and the Question of What's Still YoursJun 27
How Diffusion Models Work: Complete Guide to AI Image Generation (2026)Jun 25
Krea 2 Technical Report: Open-Weights Image Foundation Model Built for Creative ExplorationJun 23
Moebius: 0.2B Parameters, 10B-Level Inpainting, 15× Faster Than FLUXJun 23
Seedance 2.5: ByteDance's 30-Second 4K AI Video ModelJun 22
HappyHorse 1.1: Alibaba Upgrades Its Top-Ranked Video Model With Native Audio and Multi-ReferenceJun 21
Palmier Pro: The Open Source Video Editor Where Claude Edits the Timeline With YouJun 21
Voicebox: The Free, Open Source AI Voice Studio That Replaces ElevenLabs and WisprFlow in One AppJun 20
Ideogram 4.0: Open-Weight Image Generation — How to Run, API & JSON Prompts (2026)Jun 19
"Bathed in Golden Light": What Experts and the Internet Actually Think About Midjourney MedicalJun 18
Midjourney Medical: Full-Body Ultrasonic CT Scanner, 60 Seconds, No Radiation — Everything from the Official AnnouncementJun 17
Midjourney Medical: The Full-Body Scanner That Was Actually Announced (Plus the Pre-Event Speculation)Jun 16
What Is Multimodal AI? Text, Image, Audio, and Video Models ExplainedJun 13
FIFA World Cup 2026: How AI Is Running the Tournament From Kickoff to Final WhistleJun 4
Miso One: 110ms Real-Time TTS Voice Model Guide 2026
May 2026
May 31
VoxCPM2: The 2B Parameter Tokenizer-Free TTS Model That Does Voice Design, Multilingual Speech, and True-to-Life Cloning (2026)May 27
OpenCut Rewrite: Open Source Video Editor Gets Plugins, Headless Mode, MCP Server, and Multi-Platform SupportMay 26
LongCat: MIT-Licensed Talking Avatar Model Revolutionizes AI Video GenerationMay 24
Frigate NVR: The Ultimate Open-Source AI-Powered Camera System for Home Assistant in 2026May 22
Runway Aleph 2.0: Professional Video Editing vs. Google Gemini OmniMay 21
How to Remove Objects from Videos with AI: Complete Guide to Video Object Removal 2026May 20
ViMax: Agentic Video Generation - Director, Screenwriter & Producer All-in-One (2026)May 4
Runway Characters: real-time conversational video agents from one image