Open-source alternative to Vid.AI

What Vid.AI does and who it is for

Vid.AI is a hosted AI video generator that turns a typed idea into a finished video with a script, a voiceover, visuals, music and captions S1. It makes faceless YouTube videos, Shorts, TikToks and Reels S1, and the pricing page says videos can run up to 20 minutes S2. The product is pitched at creators who want "monetizable" videos, and the homepage says it is trusted by 100K+ creators S1.

The feature list on the homepage covers AI script writing, ElevenLabs voices or your own, cinematic intros, motion graphics, automatic captions, Storyblocks stock footage, background music, fake text stories, split-screen videos, animated AI avatars and timeline editing S1. The pricing page lists voice cloning as part of every plan S2. Stock footage comes from a Storyblocks library of more than 3 million licensed clips, matched to the script S1. Cinematic intros are 5 second opening shots filmed by an AI video model from your topic, with Seedance 2.5 used on the Plus and Pro plans S1. Narration uses 33 hand-picked ElevenLabs v4 narrators, or any voice from the ElevenLabs Voice Library S1.

One feature stands out for agent builders: Vid.AI exposes a Model Context Protocol (MCP) server, so you can create complete videos from a conversation in Claude or ChatGPT, or connect any AI agent over Streamable HTTP S1. Video creation through MCP draws on your subscription and credits S1.

Pricing

There is no free trial, because every video costs real compute to generate, and refunds of unused credits are offered within 30 days of payment S2. All plans include AI script writing, AI voiceovers, automatic captions, premium stock footage and music, and Animated Avatars S2.

• Starter costs $39 a month or $369 a year, with 40 one-minute short-form videos or 4 ten-minute long-form videos a month, and uses the Seedance 1 Pro Fast and Wan 2.2 video models S2.
• Plus costs $99 a month or $947 a year, with 80 short-form or 8 long-form videos a month, and adds Veo 3.1 Fast, Seedance 2.5 and Kling 2.5 Turbo Pro S2.
• Pro costs $199 a month or $1,919 a year, with 300 short-form or 30 long-form videos a month, plus priority support S2.

Yearly billing saves up to 21%, and eligible new subscribers get 50% off their first month on monthly plans S2. Everything you make draws from plan credits of 200 to 1,500 a month, and credits you top up carry over S2. Avatars have their own credit costs: creating one uses 20 credits, an intro or outro in a video uses 10 credits each, and on-screen moments cost 5 credits for every 10% of the video they cover S2. Checkout runs through Stripe S2.

What it earns

Vid.AI's monthly recurring revenue is $46,287 as of 22 October 2026, Stripe-verified on TrustMRR S3. Agent Made grades this evidence A, which means the figure comes from a Stripe connection verified by TrustMRR rather than from a founder's public claim S3.

Our take: what an open version needs

This section is opinion and analysis from Agent Made and is not a sourced fact.

Vid.AI's value is the whole pipeline rather than any single model. An open, self-hosted alternative would need to cover the same chain end to end: a script writer that works from a topic or a pasted script, text-to-speech with a choice of narrators, scene-by-scene matching of stock footage or generated images to the script, word-timed captions, a music bed, and a render step that outputs both vertical and 16:9 formats. The extras that Vid.AI sells as differentiators are generated intros from a video model, animated avatars with lip sync, templated formats such as fake text stories and split-screen, and a timeline editor for fixing scenes before export. An agent interface matters too: an MCP server that lets Claude or any other assistant drive the pipeline is now table stakes for this category.

A careful open version could also improve on the hosted product in a few places.

• Bring your own provider. Let the user plug in their own LLM, speech and video model keys instead of buying credits, so the cost of a video is the raw API price and the model choice is theirs.
• Source citations in the script. Fact-style channels (ocean facts, history explainers, space documentaries) are a large share of this genre. Keeping the sources the script was written from, and showing them in the description or on screen, would make the output more trustworthy than a script written from a prompt alone.
• Quality over volume. Hosted plans are sold by the number of videos per month. An open tool has no reason to push volume, so it can spend more compute per video on checking the script and picking better footage.
• Licensing clarity. Stock libraries like Storyblocks are the hard part to replace. An open tool should be explicit about which footage sources are licensed for commercial use and let the user swap in their own clip library.

Open tools you can use today

These are the open-source projects in the Agent Made directory that cover parts of this space.

• hypit.ai clones viral videos with AI agents into editable, repeatable workflows, open source and bring-your-own-key S5. Code: github.com/hypit-ai/hypit S5.
• Reelbench Studio is a local agent that turns a novel into a short drama, covering script, storyboard, images and video S6. Code: github.com/eternityspring/shuohao-skills S6.
• PRINTFILM is an AI short-video platform that works from a prompt and a style S7. Code: github.com/yi1108/printfilm S7.