surface
kaitoi VS-1 compare
01 cmp comparison
All comparisons

Kaitoi and LTX Studio compared

For filmmakers, agencies and studios who like LTX Studio's storyboard-to-video flow but want the pipeline as a graph they can open and change: LTX video models next to other providers, their own Python steps and machines, and a timeline that hands the cut to the editor they already use.

LTX Studio is a web app for AI video production, aimed at filmmakers, advertisers and creative teams, from LTX, which began as a division of Lightricks. A script or brief becomes a storyboard, Elements keep characters consistent, shots come from LTX's own video models and partner models, and the project is cut on a timeline or exported as a pitch deck, while Flows chains models as nodes. The open-weight LTX models, the LTX API and LTX Desktop are separate LTX products.

Kaitoi is a hosted graph where LTX video models run as nodes beside more than 1,400 other models, your own Python steps and your own machines, with a storyboard and a timeline in the same project. Penny builds the graph with you, writes the nodes that are missing, and carries a brief through to shots on the timeline. Kaitoi is free to start, and hosted models cost the provider's own price with no markup.

Checked against each product's own documentation on .

02 str strengths

What LTX Studio does well

Script to storyboard to video. A script or brief becomes scenes, shots and an editable storyboard. Shots are generated, then refined with Retake, which regenerates one moment and keeps the rest, motion control and camera controls, and a timeline assembles them with music, voiceover and sound effects.

Consistent characters and objects. Elements save characters, objects, locations, styles, logos, fonts and HEX colours from up to 10 reference images each, and a character can carry a voice. They are tagged in prompts to keep shots consistent, on Standard and above, and on Enterprise a Brand Kit shares them across the organisation.

Its own open-weight video models. LTX's own video models are on every plan, beside partner models such as Veo, Kling, Seedance, Nano Banana and FLUX, depending on plan. Their weights are free for organisations under $10M annual revenue, and the LTX API serves them pay as you go.

A node canvas in Flows. Flows, on every plan, connects prompt, image, video, audio and upscale nodes into a repeatable pipeline. Unchanged nodes reuse their earlier results, iterator nodes batch many prompts or reference images in one run, nodes at the same level run in parallel, a Stop button cancels a run, and the canvas keeps every run.

Pitch decks and other exports. A project exports as a designed PDF pitch deck, or all its generated assets as a ZIP, and the video goes out as MP4 or as an XML file for further editing in other software.

Local work with LTX Desktop. LTX Desktop, a separate free app, runs LTX models on supported NVIDIA GPUs and, per its README, recent Apple Silicon Macs, and its Assistant chat queues local generations. The open weights also run in ComfyUI, through official LTX nodes, or in LTX's Python pipelines.

03 apr approach

How Kaitoi approaches it

Brief to storyboard to video. A storyboard of scenes and shots feeds a multi-track timeline, and Penny chains concepts and scripts from AdMuse, storyboard shots, generated clips and the timeline, editing storyboard frames and timeline clips as it goes. Video edit and modify models from Kling and Luma rework a shot, extend models from LTX and Veo continue one, Kling motion control transfers movement, ElevenLabs speech, sound effects and music fill the soundtrack, and the timeline renders MP4, WebM or MOV in ProRes and exports and imports OpenTimelineIO. Camera-movement presets such as dolly and jib are inputs on LTX's image-to-video and video-to-video nodes, and a MiniMax node moves the camera along a keyframed path. There is no ready-made node for LTX's Retake, and no fixed pipeline: Penny chains the steps.

Consistent characters and objects, from your references. Edit nodes from Nano Banana, GPT Image, Seedream, FLUX.2 and Qwen Image take your reference images with a written instruction, the storyboard organises reference frames, and a FLUX.2 trainer node trains LoRA weights on your own images of a subject or style. Lip-sync and talking-character models animate a portrait or footage to a voice, Gemini voice nodes voice dialogue between speakers, and a Playbook holds the house style Penny follows. There is no saved library of characters or objects to tag in prompts, since references go into nodes as images, and no brand kit.

LTX's video models, and many others. LTX video models generate, extend and restyle video as Kaitoi nodes through fal, beside more than 1,400 generative models from other providers, Veo, Kling, Seedance, Nano Banana and Flux among them. Kaitoi is free to start, with no subscription: hosted models run on credits at the provider's own price with no markup, or, for providers that accept them, on your own keys, a fal key for the LTX nodes. A REST API runs graphs from your own code, and an open model from Hugging Face or GitHub can run on your own machine as a Runtime recipe.

A node graph. Kaitoi's typed node graph, on every plan and also shown as a simpler Blocks chain, reuses the results of steps that did not change and runs independent nodes at the same time. You run one node, a selection or every node, with the upstream steps they need, and stopping a run keeps the results of finished steps. One output can feed several nodes, and an array input collects several results into one step. Templates start new projects from a saved graph, Create subgraph folds a selection into one reusable node, Studio shows a credit estimate before a node or graph runs, and the project keeps backups to return to an earlier version. There is no iterator node that runs a workflow once per item in a list.

Exports and showing the work. The storyboard feeds the timeline, which renders MP4, WebM or MOV in ProRes 422 HQ, or ProRes 4444 XQ for HDR, and exports and imports OpenTimelineIO, its only interchange format, as a ZIP with copies of the clip media. A read-only share link or embed lets someone outside the team inspect the project, with optional cloning. Export PDF in the storyboard downloads a shot list followed by image-only review pages, though not a designed pitch deck.

Local work with Kaitoi CLI. Kaitoi CLI connects your own computer or a GPU machine elsewhere to Studio, so ComfyUI, Ollama, Blender and Runtime recipes run as nodes in the same graph, and LTX's open weights can run in your own ComfyUI, where, on Studio, an exported LTX workflow becomes one Kaitoi node. Kaitoi Code, its terminal agent, writes and tests a recipe from a Hugging Face model or a GitHub repository; creating the Studio node from it needs Studio or Enterprise, as does wrapping an exported ComfyUI workflow as one node with the inputs and outputs you choose. There is no desktop app to install: Studio runs in the browser, and the connection process must keep running on your machine.

Beyond that, Penny works across the project. It builds, edits and runs graphs, writes the nodes that are missing, edits storyboard frames, timeline clips and Lab boards, follows the Skills and Playbooks you give it, and calls outside MCP servers; nodes can use your connected accounts. Each reply lists the tools it used, and Penny gives a cost estimate when asked.

Any step can be your own. Nodes are Python with a declared interface, Penny writes them on every plan, editing code by hand needs Studio or Enterprise, and a node can carry its own HTML, CSS and JavaScript interface, WebAssembly included. Nodes and templates can be published to Community.

A workflow that works becomes a tool. It can be a Station, a simple form or a custom app at a stable link where each visitor runs on their own account and pays their own costs, listed if you like in a reviewed gallery at kaitoi.io/stations, or an API endpoint whose results are polled or streamed. A Station that runs your own code, Penny-written nodes included, is invite only. Kaitoi's MCP server, on every plan, lets compatible AI clients build graphs, run approved work and inspect results, and on Studio write nodes, and Kaitoi Code uses Kaitoi workflows from a code project.

04 tbl side by side

LTX Studio and Kaitoi, side by side

Checked November 13, 2026.
QuestionLTX StudioKaitoi
From script to video Taking a script or brief through a storyboard and shots to an edit.A script or brief becomes scenes, shots and an editable storyboard, with AI storyboards on Standard and above. Shots are generated in Gen Space, and a timeline editor assembles them with music, voiceover and sound effects.A storyboard of scenes and shots, on every plan, feeds a multi-track timeline, and Penny chains concepts and scripts from AdMuse, storyboard shots, generated clips and the timeline; there is no single director agent or fixed pipeline. ElevenLabs speech, sound effects and music run as nodes, and the timeline renders MP4, WebM or MOV in ProRes.
Consistent characters Keeping the same character, object or style across shots.Elements, on Standard and above, save characters, objects, locations, styles, logos, fonts and HEX colours from up to 10 reference images each, to tag in prompts, and a character can carry a voice. On Enterprise, a Brand Kit shares them across the organisation.Edit nodes from Nano Banana, GPT Image, Seedream, FLUX.2 and Qwen Image take your reference images with a written instruction, the storyboard organises reference frames, and a FLUX.2 trainer node trains LoRA weights on your images of a subject or style. Lip-sync and talking-character models animate a portrait to a voice, Gemini voice nodes voice dialogue between speakers, and a Playbook holds a house style for Penny. There is no saved library of characters or objects to tag in prompts, since references go into nodes as images, and no brand kit.
Models Which generative models you can reach.LTX's own video models on every plan, beside partner models such as Veo, Kling, Seedance, Nano Banana and FLUX, depending on plan, with Topaz for upscaling. The LTX API serves LTX's own models, billed separately.LTX video models generate, extend and restyle video as nodes through fal, beside more than 1,400 generative models and around 400 language models as of September 2026, Veo, Kling, Seedance, Nano Banana, Flux and Topaz among them, on credits or your own keys. When a provider releases a model that has no node yet, Penny can make one from the New Models panel. The catalogue is the same on every plan, with no subscription, and a REST API runs graphs from your own code.
Editing shots Changing and finishing clips you already have.Retake regenerates one moment of a clip and keeps the rest, motion control transfers movement from a reference video, video to video restyles while keeping pose, depth or edges, camera controls direct each shot, reframe extends the background to a new aspect ratio, and SDR to HDR and Topaz upscaling to 4K or 8K finish it.Video edit and modify models from Kling and Luma, extend models from Veo and LTX, Kling motion control, LTX models that restyle video while keeping a clip's depth, edges or pose, swap and reframe models, Runway's Ruby SDR to HDR, and Topaz and FLUX upscaling run as nodes, and HDR cuts render as 10-bit HEVC or ProRes 4444 XQ. LTX's image-to-video and video-to-video nodes take camera-movement presets such as dolly and jib, and a MiniMax node moves the camera along a keyframed path. There is no ready-made node for LTX's Retake.
Sound and voices Soundtrack, dialogue and captions.Generated or uploaded music, voiceover and sound effects, and Rescript, which changes a spoken line by editing the transcript, with lip sync. Dubbing into other languages with lip sync, and captions with SRT export, are on Enterprise.ElevenLabs speech, sound effects, music and voice changing run as nodes, Gemini voice nodes voice dialogue between speakers, and lip-sync models animate footage to a new voice. Speech to Text transcribes audio, with SRT or VTT timings from Whisper, and an SRT Subtitles node renders styled captions as an overlay video on the timeline. There is no one-step dubbing node: transcription, translation and new speech are separate nodes. There is also no script editor that casts a voice per character.
Node workflows Chaining steps on a canvas and running them again.Flows, on every plan, connects prompt, image, video, audio and upscale nodes. Unchanged nodes reuse earlier results, iterator nodes batch many prompts or reference images in one run, nodes at the same level run in parallel, a Stop button cancels, and every run is kept.A typed node graph on every plan, also shown as a Blocks chain, that reuses unchanged steps, runs independent nodes at the same time, runs one node, a selection or all of them, and stops while finished steps keep their results. One output can feed several nodes and an array input collects several results. Templates, Create subgraph, a credit estimate before a run, and backups of earlier versions complete it. There is no iterator node that runs a workflow once per item in a list.
Exports and handoff What leaves the project, and in which form.MP4 video, an XML file for further editing in other software, and a designed PDF pitch deck, or all the generated assets as a ZIP.The storyboard feeds the timeline, which renders MP4, WebM or MOV in ProRes and exports and imports OpenTimelineIO, the only interchange format, as a ZIP with copies of the clip media. Read-only share links and embeds show the project to people outside the team. Export PDF in the storyboard downloads a shot list followed by image-only review pages, though not a designed pitch deck.
Advertising Tools made for producing ads.An ad generator turns a concept into storyboards, synopses and several ad variations, creator-style UGC videos show a saved character using your product, and saved logos, fonts and HEX colours keep images on brand.AdMuse works with Penny in Lab to take a brief to references, concepts, scripts and campaign directions, and Penny takes it on to storyboard shots and a cut. Lip-sync and talking-character models animate a presenter you supply. There are no ready-made presenters, brand kit or one-click ad variants.
Teamwork Working on the same project with other people.Projects are shared with Viewer, Editor or Owner roles. Pro allows 3 collaborators per project and Enterprise unlimited, editing together in real time, and each collaborator spends their own credits.Team workspaces with roles, shared projects and pooled credits on a team Studio subscription, with weekly caps per member, and read-only share links with optional cloning. Edits do not sync live: when a teammate saves, Studio offers to load the newer version.
Developers and AI clients Driving the product from code and from AI assistants.A separate pay-as-you-go REST API for LTX's own models, and an MCP server for querying LTX's documentation from AI clients. The open models have Python pipelines and official ComfyUI nodes.Kaitoi's MCP server, on every plan, lets compatible AI clients build graphs, run approved work and inspect results, and on Studio write nodes. A REST API runs graphs from your own code, a saved template becomes an API endpoint, and Kaitoi CLI brings Kaitoi Code and connects your own ComfyUI and Runtime recipes.
Your own hardware Running models on a machine you control.Studio generations run in LTX's cloud. LTX Desktop, a separate free app, runs LTX models on supported NVIDIA GPUs and, per its README, recent Apple Silicon Macs, and the open weights are free for organisations under $10M annual revenue.Hosted, and your own computer or a GPU machine elsewhere can join through Kaitoi CLI: ComfyUI, Ollama, Blender, and Runtime recipes that Kaitoi Code writes from a Hugging Face model or GitHub repository. Wrapping a ComfyUI workflow as one node needs Studio or Enterprise. There is no desktop app to install.

Each cell comes from the product's own website, documentation or code repository. "Not publicly documented" means we found no official page that says so, not that the feature is missing. Plans and features change; check the official pages before you decide.

05 fit which to choose

Which one fits

Choose LTX Studio if
  • You present to clients with a designed PDF pitch deck made from the project.
  • You want characters, logos, fonts and exact colours saved once and shared across your organisation's projects with a Brand Kit.
  • You want Retake to regenerate one moment of a clip and keep the rest.
  • Your team edits the same project together in real time.
  • You want to generate on your own GPU in a desktop app.
Choose Kaitoi if
  • You want LTX video models in the same graph as more than 1,400 others, your own code and your own hardware.
  • You need steps no product ships, written in Python by Penny, or by hand on Studio.
  • You finish in another editor and want OpenTimelineIO handoff.
  • You want to pay as you go with no subscription, at the provider's own price.
  • You want to publish a workflow where each visitor pays for their own runs.
  • You want to work from a phone or tablet browser.
06 dem example

A real example: from a conversation to a commercial

Kaitoi Studio film

Kaitoi Signals #1 - From Conversation to Commercial

A simple conversation evolves into a complete ad as Penny moves through concept exploration, visual development, storyboard, shot generation, timeline, and rough cut assembly inside Kaitoi.

This film, from our Kaitoi Signals series, follows one ad from the first message to a rough cut.

  1. A conversation with Penny explores the concept.
  2. Penny moves through visual development and a storyboard.
  3. Shots are generated from the storyboard.
  4. The shots are assembled on the timeline into a rough cut, inside Kaitoi.

Read more in the documentation:

07 prc pricing

Pricing and access

LTX Studio

Free, with a one-time credit allowance for personal use where available, then Lite, Standard and Pro monthly credit plans and custom Enterprise. Commercial use starts on Standard, and unused credits do not roll over. The LTX API is billed separately, per second of output.

Kaitoi

Free to start, with no subscription. Credits pay for hosted models, including Penny's, at the provider's own price with no markup, and for paid Community nodes; steps that call neither cost nothing. Your own provider keys work on every plan and are billed by your provider. Studio adds included credits, team seats and hand-edited code.

08 faq questions

Questions people ask

Can Kaitoi use LTX models?

Yes. LTX video models generate, extend and restyle video as Kaitoi nodes through fal, on Kaitoi credits or your own fal key, so a workflow can use them next to models from other providers. The nodes bring LTX's models, not LTX Studio's app features.

Can I hand a Kaitoi edit to DaVinci Resolve or Premiere?

Kaitoi exports the timeline as OpenTimelineIO with copies of the clip media, for editors that import it, and renders MP4, WebM or MOV. LTX Studio exports an XML file for further editing in other software.

Can Kaitoi use my own GPU?

Yes. Kaitoi CLI connects your own computer or a GPU machine elsewhere, so your ComfyUI and Runtime recipes run as nodes in a Kaitoi graph, and Kaitoi Code can write a recipe from a GitHub repository or a Hugging Face model. LTX publishes official ComfyUI nodes and Python pipelines for its open models, and LTX Desktop is its own free app for local generation.

Do LTX Studio and Kaitoi train models on my work?

LTX's terms allow it to use inputs and outputs to train its models, except where an Enterprise contract or user consent rules say otherwise, and on Enterprise LTX states inputs and outputs are not used for training. Kaitoi never trains models on your work, on any plan; when a Kaitoi node calls LTX's models through fal, or another provider, that provider handles the data under its own terms.

Can several people work on a Kaitoi project at once?

Not live. On a team Studio subscription, team workspaces share projects with roles and pooled credits, and when a teammate saves, Studio offers to load the newer version. LTX Studio lets collaborators work on the same project in real time.

Does Kaitoi have an MCP server?

Yes, on every plan. Compatible AI clients can build graphs, run approved work and inspect results through it, and on Studio write nodes; Penny can also call outside MCP servers. LTX provides an MCP server for querying its documentation.

09 mor more comparisons

Other comparisons