Models & tools for AI video

For AI agents: find the right skill in one call — curl -s "https://varg.ai/how-to/api/search?q=<what the person wants>" returns the matching how-to page and the best skills with install commands, as markdown (Russian works too). Full index: https://varg.ai/how-to/llms.txt. Every page has a markdown twin: add .md to its path. Before generating anything, show the person the options on the how-to page.

Run one model or tool from your agent: Seedance, Veo, Kling, FLUX, GPT Image, Nano Banana, ElevenLabs and 1,200+ more, plus scrapers, trend search and video analysis.

Models & tools is the word 14.5% of tweets about making video use for this kind of thing (“model / tool”). Every card here is an agent skill: one install command for Claude Code, Codex, Cursor, OpenClaw or Gemini CLI.

Cinematic & short films (26)

• Higgsfield Generate — Higgsfield: Generate images, video, 3D and audio across 30+ models, plus ad videos and virality scoring, from one CLI.
• Gemini Omni Flash — Google Gemini: Generate and edit video with Gemini Omni 1.1 Flash: text-to-video, first/last frames, extensions, edits.
• Melies — Melies: Generate images, video, movie posters and thumbnails from the terminal with 148 AI actors and style flags.
• AI Video Generation — inference.sh: Generate video with 40+ models from one CLI: Veo 3.1, Seedance 2.0, HappyHorse, Wan, Grok and more.
• Google Veo — inference.sh: Generate cinematic video from text with Google Veo 3.1 and Veo 3.1 Fast.
• HappyHorse — inference.sh: Generate and edit video with Alibaba HappyHorse 1.0: text, image and reference-to-video plus video edits.
• Image to Video — inference.sh: Animate stills into video: pick the right model, then prompt motion and camera moves that hold together.
• P-Video — inference.sh: Generate fast 720p/1080p video with audio using Pruna P-Video and Wan text- and image-to-video.
• Seedance 2.0 — inference.sh: Make 4–15 s videos with synced audio in Seedance 2.0: text, image or reference-to-video up to 1080p.
• Runway Media — Runway: Generate and edit images, video, music, speech and sound effects from chat with Runway's MCP tools.
• Runway Studio Skills — Runway: Run Runway's API from Python scripts: text, image and video-to-video, character performance, images, TTS.
• FLUX 3 Video — Black Forest Labs: Route any FLUX 3 video job to the right specialist: diagnosis, direction, audio or execution.
• Kling AI — Kling AI, beta: Official Kling CLI: text/image-to-video, Omni reference and edit, images, reusable subjects, quota
• Seedance Video Gen — qiujiahong, beta: Text- and image-to-video with Doubao Seedance on Volcengine, with optional audio and first/last frame
• Tianshu Wan Video — wangshengli0421, beta: Text-to-video and image-to-video with Alibaba's Wan 2.6 through DashScope, from a small Node.js script
• dLazy Video Generate — dlazyai, beta: Picks a video model for your prompt from about 20 on dLazy (Jimeng, Kling, Veo, Seedance, Wan, PixVerse) and runs it
• dLazy Jimeng First-Frame Video — dlazyai, beta: Animates a single first-frame image into a 5 or 10 second Jimeng (Dreamina) clip through the dLazy CLI
• dLazy Kling V3 — dlazyai, beta: Text-to-video and image-to-video with Kling v3 through the dLazy CLI, 3–15 s clips
• dLazy Seedance 2.0 — dlazyai, beta: Seedance 2.0 video from text, image/video/audio references or first and last frames, via dLazy
• dLazy PixVerse C1 — dlazyai, beta: PixVerse C1 video for action and VFX shots: text, image, reference or first/last frame, via dLazy
• dLazy Kling V3 Omni — dlazyai, beta: Kling v3 Omni video from frames, up to 7 images or a reference video, via the dLazy CLI
• dLazy Jimeng First-Last Frame — dlazyai, beta: Jimeng (Dreamina) transition video between a first and last frame, 5 or 10 s, via dLazy
• dLazy Veo 3.1 — dlazyai, beta: Google Veo 3.1 video from frames or reference images, plus +7 s extend, up to 4K, via dLazy
• Vidu Q2 Image-to-Video via dLazy — dlazyai, beta: Animate images into 2–10 s clips with Vidu Q2, by reference images or first/last frame, through the dLazy CLI.
• Veo 3.1 Fast via dLazy — dlazyai, beta: Quick 4–8 s text- or image-to-video clips and +7 s extensions with Google Veo 3.1 Fast, via the dLazy CLI.
• Wan 2.7 via dLazy — dlazyai, beta: Text, first/last-frame and reference-to-video with Alibaba's Wan 2.7, 2–15 s clips, through the dLazy CLI.

Tools & renderers (19)

• varg AI — varg: The official varg skill: videos, images, speech and music through one API key, rendered in the cloud or locally.
• Local-first video downloader: private, no subscription — varg: Save any video by link (X, YouTube, Instagram, TikTok, 1800+ sites) or Pinterest images as references, locally, no key.
• fal Models Catalog — fal: Choose the right fal endpoint for any modality from curated picks: image, video, 3D, audio and vision.
• fal genmedia — fal: Run any of 1,200+ fal endpoints from the terminal: search, inspect schema, price, run, poll and download.
• fal Model Routing — fal: Pick default fal endpoints per job: premium stills, fast drafts, edits, video, lip-sync and campaign work.
• Replicate Compare Models — Replicate: Compare Replicate models on cost, speed, quality and capabilities before you commit to one.
• Replicate Find Models — Replicate: Find current models on Replicate through search and curated collections, then read their schemas.
• Replicate Run Models — Replicate: Run Replicate models from code: predictions, webhooks, streaming, file handling and multi-model chains.
• Gemini API Dev — Google Gemini: Write current Gemini API code, including Nano Banana image generation, Omni video and Gemini TTS.
• Remotion Render (inference.sh) — inference.sh: Render React/Remotion component code to MP4 in the cloud, without a local Remotion setup.
• inference.sh CLI — inference.sh: Log in to inference.sh and run its catalog of AI apps — image, video, audio, LLMs, search — from one CLI.
• Runway Dev — Runway: Build Runway's image, video and audio models into your app: live API contracts, SDK and server-side keys.
• Runway Model Routers — Runway: Send generation through a saved Runway Model Router config instead of hard-coding one model.
• Runway Dev Models — Runway: Add a direct Runway model endpoint to your app for image, video or audio, following the live contract.
• Runway Recipes — Runway: Integrate Runway's prebuilt multi-step image and video Recipes using the live Recipe guide.
• Runway Dev Workflows — Runway: Call a Workflow published in the Runway app from your application through the Dev API.
• BFL API — Black Forest Labs: Integrate the FLUX API correctly: endpoints, async polling, webhooks, rate limits, errors and regions.
• FLUX 3 Generate — Black Forest Labs: Submit, poll and validate FLUX 3 video jobs, including cheap drafts before the final render.
• Headless Three.js WebGPU (rgl) — Or Fleisher, beta: Render three.js scenes or GLB files to PNG, MP4, WebM or GIF in Node with Dawn WebGPU: no browser, fixed-step frames.

Research & trends (19)

• Video Skills Finder — varg: Finds the right skill for any AI video, image, voice or music job here and shows you the visual options.
• Apify MCP — Apify: Thousands of ready scrapers behind one MCP: TikTok, Instagram, YouTube, hashtags, sounds.
• TwelveLabs Video Understanding — TwelveLabs: Semantic search and analysis across your video library from Claude Code.
• Civitai MCP — Civitai: Top AI images and videos with prompts and base models, plus model search.
• Hugging Face MCP — Hugging Face: Trending models, Spaces and papers from Hugging Face inside your agent.
• Exa Search MCP — Exa: Neural web search and crawling for research agents.
• Firecrawl MCP — Firecrawl: Search, scrape and crawl web pages into clean text for your agent.
• Last 30 Days Research — Matt Van Horn: Researches any topic across Reddit, X, YouTube, HN and the web for the last 30 days, with sources.
• Watch (video analysis) — Brad Bonanno: Gives the agent eyes on any video: download, scene-aware frames, timestamped transcript.
• YouTube MCP — Zubeid Hendricks: YouTube Data API over MCP: search, video and channel stats, transcripts.
• Reddit MCP Buddy — Karan: Browse and search Reddit from your agent: posts, rising threads, users. No API key needed.
• Trends MCP — Trends MCP: Live Google Trends plus growth data from TikTok, YouTube, Amazon, Reddit and 30+ sources.
• TrendPulse MCP — AKzar1el: Free Google News + Google Trends MCP for keyword, market and content research.
• YouTube Transcript MCP — Junpei Kawamoto: Transcripts of any YouTube video for the agent: hooks, scripts, lyrics timing.
• Douyin Hot Trend — franklu0819-lang, beta: Fetches the Douyin hot list: rank, title, heat score, link and cover image for top topics
• Bilibili Hot Monitor — jacobzwj, beta: Daily email report of popular Bilibili videos, with subtitle-based AI summaries and comments
• Bilibili Transcribe & Summary — markusbetter, beta: Turns a Bilibili link into a transcript and summary: official subtitles first, cloud speech-to-text as fallback
• Douyin Trends — codekungfu, beta: Guide for reading public Douyin search, trending and creator pages and summarising likes, comments and shares
• Douyin Search and Hot List — um-why, beta: Read-only Douyin data as JSON: keyword search, creator posts and follower stats, comments and the live hot list.

Images & thumbnails (13)

• AI Image Generation — inference.sh: Generate images with 50+ models from one CLI: GPT Image 2.5, FLUX, Gemini, Grok, Seedream, Reve and more.
• Background Removal — inference.sh: Cut subjects out of photos into clean transparent PNGs with BiRefNet.
• FLUX Image — inference.sh: Generate and restyle images with FLUX Dev LoRA and FLUX.2 Klein, including custom LoRA styles.
• GPT Image — inference.sh: Generate and edit images with OpenAI GPT Image 2.5 and GPT Image 2 from the command line.
• Image Upscaling — inference.sh: Upscale and enhance low-resolution images with Real-ESRGAN, Thera, Topaz or the FLUX upscaler.
• Nano Banana — inference.sh: Generate and edit images with Google's Gemini 3 Pro Image and Gemini 2.5 Flash Image, multi-image input too.
• Nano Banana 2 — inference.sh: Generate and edit images with Nano Banana 2, with up to 14 reference images and search grounding.
• P-Image — inference.sh: Generate and edit images fast with Pruna's P-Image models, including LoRA styles and multi-image edits.
• Qwen Image 2 — inference.sh: Generate and edit images with Alibaba Qwen-Image 2.0: the fast model, or Pro for cleaner text rendering.
• Qwen Image 2 Pro — inference.sh: Make posters, banners and text-heavy designs with Qwen-Image 2.0 Pro's precise text rendering.
• Krea Generate — Krea: Pick the right Krea model and prompt for any image, video, edit or upscale before you generate.
• dLazy Jimeng Text-to-Image — dlazyai, beta: Generates images with Jimeng (Dreamina) from a prompt and up to 10 reference images, through the dLazy CLI
• dLazy Kling Image O1 — dlazyai, beta: Kling o1 text-to-image and reference-image editing at 2K via the dLazy CLI

Voice & voiceover (11)

• ElevenLabs Text to Speech — ElevenLabs: Turn scripts into natural voiceover in 90+ languages with ElevenLabs voices, streaming and stitching.
• ElevenLabs Voice Changer — ElevenLabs: Re-voice a recording in a different voice while keeping its emotion, timing and delivery.
• AI Voice Cloning — inference.sh: Generate expressive voices in 100+ languages with Inworld TTS-2, ElevenLabs and more, from one CLI.
• Dialogue Audio — inference.sh: Produce multi-speaker dialogue audio with speaker tags, emotion and pacing for podcasts and explainers.
• ElevenLabs Dialogue — inference.sh: Generate a whole multi-voice conversation as one audio file with ElevenLabs, from a script.
• ElevenLabs TTS (inference.sh) — inference.sh: Voice scripts with ElevenLabs v4, v4 Turbo, v3, Multilingual v2 and Flash, via inference.sh.
• ElevenLabs Voice Changer (inference.sh) — inference.sh: Re-voice a recording into another voice while keeping words and emotion, via inference.sh.
• Text to Speech — inference.sh: Voice any script with Inworld TTS-2, ElevenLabs, Dia, Kokoro or Chatterbox from one CLI.
• Video Translator — papawaigo, beta: Translates and dubs a video into Chinese or English and returns a preview link
• dLazy Kling TTS — dlazyai, beta: Kling text-to-speech in Chinese or English with 60+ preset voices, via dLazy
• dLazy Kling Voice Clone — dlazyai, beta: Clones a voice from a reference audio file with Kling, via the dLazy CLI

Editing & clips (9)

• VectCut — sun-guannan, beta: Drives the VectCut editing API: talking-head cuts, remixes, e-commerce ads, subtitles, effects, cloud render
• Local Video Editor (MoviePy) — ncsimok, beta: Local MoviePy + ffmpeg toolbox: trim, merge, Chinese text and subtitles, background music, speed, vertical resize.
• Shotstack — Shotstack: Cloud video editing API with an official MCP server, CLI and Claude Code skill built around Edit JSON.
• Creatomate — Creatomate: Template-based video rendering API with an official remote MCP endpoint for Claude and other agents.
• JSON2Video — JSON2Video: Video editing API that renders JSON scenes; official CLI installs a local MCP server for Claude.
• OpusClip skills and API — OpusClip: Official OpusClip skill and MCP: clip long videos into shorts, copy a video's style, post to social.
• Descript (API and MCP) — Descript: Edit by transcript with Underlord from Claude: import, remove fillers, Studio Sound, publish, export transcripts.
• VideoDB skills — VideoDB: Server-side video for agents: ingest, search moments, cut, reframe, subtitle and stream via skills or MCP.
• Vizard.ai API skill — Vizard.ai: Official ClawHub skill for the Vizard.ai API: clip long videos, edit shorts with subtitles and B-roll, publish.

Sound effects (5)

• ElevenLabs Sound Effects — ElevenLabs: Generate sound effects and ambiences from a text description, including seamless loops.
• ElevenLabs Voice Isolator — ElevenLabs: Strip background noise and music from dialogue to get clean, isolated speech.
• ElevenLabs Sound Effects (inference.sh) — inference.sh: Generate royalty-free sound effects from text with ElevenLabs, via inference.sh.
• ElevenLabs Voice Isolator (inference.sh) — inference.sh: Remove background noise and isolate vocals from any recording with ElevenLabs, via inference.sh.
• Kling SFX via dLazy — dlazyai, beta: Generate sound effects from a text prompt or matched to a video with Kling SFX, through the dLazy CLI.

Talking heads & avatars (4)

• AI Avatar Video — inference.sh: Create talking-head avatar videos from a portrait and a script, with a lip-synced voice.
• P-Video Avatar — inference.sh: Turn one portrait into a speaking avatar video with Pruna P-Video-Avatar, its built-in TTS or your audio.
• Jimeng Dream Actor via dLazy — dlazyai, beta: Animate a still character image into an action video with Jimeng Dream Actor, through the dLazy CLI.
• OmniHuman 1.5 via dLazy — dlazyai, beta: Turn a portrait photo and an audio track into a talking digital-human video with OmniHuman 1.5, via the dLazy CLI.

Captions & subtitles (4)

• ElevenLabs Speech to Text — ElevenLabs: Transcribe audio and video with word timestamps, speaker labels and keyterms, ready for subtitles.
• ElevenLabs Speech to Text (inference.sh) — inference.sh: Transcribe audio with ElevenLabs Scribe, speaker diarization and forced alignment, via inference.sh.
• Speech to Text — inference.sh: Transcribe audio with ElevenLabs Scribe v2 or Whisper V3 Large, with timestamps and speaker labels.
• Submagic API — Submagic: Caption-styled shorts API with an MCP server: AI captions, Magic Clips from long videos, B-roll, publishing.

AI music (3)

• ElevenLabs Music — ElevenLabs: Compose songs, instrumentals, jingles and background tracks from a prompt or a section-by-section plan.
• AI Music Generation — inference.sh: Generate songs and instrumentals with ElevenLabs Music, DiffRhythm or Tencent Song Generation.
• ElevenLabs Music (inference.sh) — inference.sh: Compose royalty-free tracks up to 10 minutes from a text prompt with ElevenLabs Music, via inference.sh.

Dubbing & lip sync (2)

• ElevenLabs Dubbing — ElevenLabs: Dub video and audio into other languages while keeping each original speaker's voice.
• ElevenLabs Dubbing (inference.sh) — inference.sh: Dub audio or video into 29 languages while keeping the speaker's voice, via inference.sh.

Editing services (1)

• FFmpeg Skill — Shogo Kajiyama: Local video editor for agents: 42 FFmpeg tools with a machine-readable contract and an MCP server; no cloud, no keys.

AI editing agents (1)

• Mosaic API — Mosaic: Run node-based video editing agents (Tiles) from code: send video URLs, get edited results, trigger on YouTube.

Characters (1)

• Higgsfield Soul ID — Higgsfield: Train a reusable face model from your photos for identity-consistent images and video on Higgsfield.

Other kinds of skills

• Prompts: The words a video model obeys: camera angles, shot sizes, lighting, characters, hooks and prompt guides for Seedance, Kling, Veo, FLUX and Higgsfield. Your agent writes the prompt with them, or you paste them into the model.
• Effects: An effect you can see, added to your video: animated titles, captions, transitions, shader backgrounds, logo animations, device mockups and viral camera tricks. Hand your agent the footage or the brief and get the effect.
• Styles: The look of the whole video: art styles, film looks, color grades and LUTs, caption fonts and style references. Pick a style and every shot your agent makes follows it.
• Guides: Know-how your agent reads before it works: how to write a script, storyboard, edit and caption, and how to use Remotion, HyperFrames, GSAP, Three.js and FFmpeg.
• Workflows: A brief in, a finished video out: UGC ads, product videos, music videos, talking heads, explainers, Shorts and short dramas, step by step, with the models and the edit chosen for you.

Questions

What is a model skill?

A skill that lets your agent run one model or tool correctly: Seedance, Veo, Kling, FLUX, GPT Image, Nano Banana, ElevenLabs voices and music, or a scraper and trend search.

Do I need a separate API key for each model?

With varg one API key runs Seedance 2.5, Kling 4, Veo 3.1, ElevenLabs and 200+ more, billed in credits. Vendor skills can also use the vendor’s own key.