Hypit
Hypit is an open-source video programming language and runtime for AI agents. A coding agent installs the skill, reads a reference video or a written brief, and rebuilds the shots, dialogue, B-roll, captions and effects with your material, producing many variants per run. Hosted generation runs on credit plans, and your own model keys can be used instead.
What is Hypit?
Hypit describes itself as a video programming language, creative toolkit and runtime system for AI agents, and its homepage summarises the pitch as cloning any viral video with AI agents. The open-source core lives on GitHub, installs into a coding agent with one command, and gives that agent the production knowledge and executable tools to read a reference video, generate the material and compose an editable result. The company positions it against established ad and clip tools on three points: it adapts the whole video rather than only the script, it can produce many variants in a single run, and the core is open source, so it can be forked, modified or self-hosted.
The mechanism that separates Hypit from ordinary editors is the script format. Video projects are written as markup in which segments, speakers and phrases are named, and cuts, B-roll, graphics and effects are attached to words rather than to timecodes. A ranking graphic can be bound to the moment a name is spoken, and a B-roll clip to a specific phrase, so timing changes do not break the edit. The homepage demonstrates the format with a script carrying dialogue lines, speech-level markers and B-roll items hooked to individual phrases.
In practice a user brings a reference video or a description and tells the agent what should change: the presenter, the product, the language, the aspect ratio or the call to action. The agent studies what makes the reference work, adapts the script and the visual direction, and keeps its decisions in the project. Before paid generation starts it explains the account, the planned work and the available rates or estimated cost, so a budget can be agreed; generation then runs while the agent builds graphics and captions in parallel, and the output is kept in Build Results. Finished projects stay as ordinary editable files and can be opened in Studio for timeline edits and supported property changes.
Setup is documented in a quickstart written for agent users. The requirements are a coding agent that can use skills, such as Claude Code or Codex, plus a reference video or a description of the video. One install command adds the skill and the executable, and the agent checks for an existing installation and prepares anything missing. Speech recognition can run locally through WhisperX when captions need word timings, or the same capability can be used through a hosted account, and the project publishes a set of free AI avatars with distinct voices for generated videos.
Hypit is sold in two layers. The framework itself is free to use, and the repository, documentation, template formats such as podcast, street interview, dialogue and video call, and a Discord community are open to anyone. Hosted generation runs on credit subscriptions, which the terms of use describe alongside the company operating the service, Shenzhen Kaxi Yongliu Technology Co., Ltd. Model services can also be supplied by the user through their own API keys under a bring-your-own-key arrangement.
Hypit Core Features
Word-anchored editing
Cuts, dialogue, B-roll, captions and effects are bound to words in the script rather than to timecodes, so edits survive timing changes.
Full-video cloning
A reference video is analysed and rebuilt with your own content, shots, pacing, captions and effects included, not just the script.
Agent-native workflow
The skill installs into Claude Code, Codex or another coding agent, which plans the work, generates the material and assembles the video.
Batch variant generation
One run can produce 10, 50 or more variants of the same concept for testing.
Open-source core
The repository can be forked, modified and self-hosted, with the executable and the skill published as separate update channels.
Template formats
Proven structures such as podcast, street interview, dialogue and video call are available as starting points.
Bring your own keys
Image, video, voice and language models can be supplied through your own API keys, or used as hosted generation on a Hypit account.
Editable project files and Studio
Projects stay as ordinary editable files and can be opened in Studio for timeline edits.
Who is Hypit for?
Hypit is aimed at people who already produce video with AI-assisted tooling and who work from coding agents rather than studio software. The quickstart assumes an agent that can use skills, naming Claude Code and Codex, so the closest fit is a developer, growth engineer or technical marketer who is comfortable installing a command-line package, handing an agent a reference file and approving a generation budget. Non-technical creators can use the hosted services through an account, but the free open-source route requires an agent. The workflow suits teams that test many creative variants. The homepage claims one command, a hundred variants, and the comparison table lists batch runs of 10, 50 or more per run as a built-in capability. That fits performance marketing, where a proven ad format is cloned with different products, presenters or hooks, and it fits agencies that need the same format produced for several clients from one reference. Because Hypit can ingest a reference video and rebuild its structure, it also suits teams that localise content. A video can be re-created with a different language, presenter or aspect ratio while keeping the pacing and graphics that made the original work, which is normally split between a translator, an editor and a motion designer. Studios running several channels at once are addressed by the Ultra plan and its 27,000 monthly credits, and the terms of use allow organisations to accept them on behalf of their users. The homepage lists logos it says represent teams using the product, including Google, Microsoft, Meta, NVIDIA, OpenAI, Anthropic, Adobe, Alibaba, ByteDance, Tencent, Baidu, DiDi and two Chinese universities, which suggests the target also includes teams inside larger media or platform organisations rather than only individual creators. It is a poor fit for anyone who wants a finished video without touching a terminal or an agent, for editors who prefer a conventional timeline-first interface, and for users who need a fixed price per video before starting, since hosted generations are charged in credits that vary with the model and resolution chosen.
Hypit Use Cases
Clone a viral video format with your own product, presenter and language.
Generate dozens of ad variants from one concept and test which performs.
Turn a studio opening, ranking format or reaction clip into your own version.
Start a podcast, street interview or video-call format from a template.
Rebuild a competitor's ad structure with your own brand assets and claims.
Localise a video into another language with new dialogue, voice and captions.
Assemble B-roll, captions and motion graphics automatically from a script.
Self-host the open-source core inside your own pipeline to control costs and data.
Hypit Pros and Cons
Pros
- The open-source core can be self-hosted at no licence cost, with hosted plans needed only for generation credits.
- Agent-first design works with Claude Code, Codex and other coding agents instead of a closed studio interface.
- Model freedom: bring your own API keys for image, video, voice and language models, or use Hypit's hosted models.
- Batch output of 10, 50 or more variants per run is a built-in capability rather than an add-on.
- The quickstart states the account, planned work and estimated cost before paid generation runs, so budgets are agreed in advance.
Cons
- The free open-source path assumes a coding agent such as Claude Code or Codex and your own model provider accounts.
- Installing the skill or the executable includes no generation credits, so footage is billed either to a Hypit plan or to your own API keys.
- Reference-video cloning carries legal responsibility: the terms put the duty to clear rights for reference material, likenesses and voices on the user.
FAQ About Hypit
Hypit Pricing
Hypit's framework is free to use and open source. The hosted service sells credit subscriptions listed with yearly billing: Starter at $9.90 a month for 1,800 credits, Pro at $39.90 a month for 10,000 credits and Ultra at $99.90 a month for 27,000 credits, with the Pro and Ultra rates shown as discounts from $49.90 and $129.90. One-off top-up packs run from $15 for 2,000 credits to $130 for 20,000, and teams can request a custom plan.
Check official pricingStarter
1,800 credits refreshed every month, $118.80 billed yearly. For trying the workflow on a couple of videos; the site estimates 260 images or 38 Seedance mini clips from the monthly credits.
Pro
10,000 credits refreshed every month, $478.80 billed yearly, shown as 20% off the $49.90 rate. For creators shipping to every platform weekly.
Ultra
27,000 credits refreshed every month, $1,198.80 billed yearly, shown as 23% off the $129.90 rate. For studios running several channels at once.
Credit top-ups
Pay-as-you-go packs: 2,000 credits $15, 5,000 credits $35, 10,000 credits $65 and 20,000 credits $130. No subscription needed, top-up credits never expire, and plan credits reset each period and are spent first.
Business (Custom Plan)
Custom plan for teams and organisations with advanced needs, priced on request by email at official@hypit.ai.
Hypit Alternatives
Fetra
Fetra AI is a web-based creative suite that combines a conversational AI Director, a multi-model video and image Create studio, Viral Remix templates, an AI Avatar library, free text-to-image and image-to-image generation, an asset library, and social publishing. It brings leading models such as GPT Image 2, Nano Banana 2, Seedance, Veo, and Kling into one credit-based workspace for creators, brands, and teams.
Photo to Video AI
Photo to Video AI helps creators turn photos into polished AI videos with motion, transitions, and style control.
AI Seedance - Seedance 2.5 Video Generator
Generate directed Seedance 2.5 videos with text-to-video, image-to-video, reference guidance, synchronized audio, multi-shot storytelling, and up to 4K output.