# Is Higgsfield open source? What its GitHub repositories actually let you run No. Higgsfield's apps and models are proprietary. An audit of every public Higgsfield repo and the open-higgsfield studio (October 2026): licence, owner, what runs, what needs a key. Author: Ravi Parikh Canonical: https://voyager.so/blog/is-higgsfield-open-source Published: 2026-10-20T10:45:00-07:00 Updated: 2026-10-20T10:45:00-07:00 ![The wide-trace/open-higgsfield repository on GitHub, signed out on 20 October 2026: 3 commits by wide-trace, folders public, scripts and src, files including .env.example and package.json, and an About sidebar listing openhiggsfield.ai, 4.5k stars and 961 forks with no licence shown.](https://cdn.moda.app/voyager/blog/is-higgsfield-open-source/hero-9fcbc050ec2735b6.webp) No. Higgsfield's platform, its apps such as Cinema Studio and Genjutsu, and its own models are proprietary, and its help centre says so: "No, there is no open-source version of Higgsfield AI" ([official Higgsfield platforms](https://higgsfield.ai/creator-hub/help-center/getting-started/official-higgsfield-platforms)). The code you can find on GitHub falls into four groups, and none of them generates an image or video on your own machine: * **Official client tools** (SDKs, CLI, agent skills, a Cursor plugin), which call Higgsfield's servers and bill your account. * **An official studio template**, a web interface for Higgsfield's paid API, published without a licence. * **open-higgsfield**, the unlicensed studio a Higgsfield-affiliated personal account linked in September after saying "we also open sourced everything we've built over the last 18 months", built to work with the same paid API. * **A 2023 framework for training language models**, open source and Higgsfield's own, with nothing to do with the video product. One naming trap first: "Open Higgsfield" is the name of Higgsfield's official API console at `open.higgsfield.ai`. The open-higgsfield repository and the sites at openhiggsfield.ai, openhiggsfield.com and thehiggsfield.ai are not among the official surfaces Higgsfield lists. Every repository below was read through the GitHub API on 20 October 2026. Higgsfield's statements come from its help centre, its blog and posts on X, linked where they appear. Voyager, the desktop app our team builds, generates with many of the same models Higgsfield offers and competes with it for some of the same work. ## What you can build with today | If you want to | Use | Licence | Runs where | Pays from | | ------------------------------------------ | --------------------------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------- | --------------------------------------------- | ------------------------------------------- | | Call Higgsfield models from your own code | [higgsfield-client](https://github.com/higgsfield-ai/higgsfield-client) (Python) or [higgsfield-js](https://github.com/higgsfield-ai/higgsfield-js) | Apache 2.0; MIT in `package.json` | Higgsfield's servers | API balance, in dollars | | Generate from a terminal or a coding agent | [CLI](https://github.com/higgsfield-ai/cli) plus [skills](https://github.com/higgsfield-ai/skills) | MIT; the CLI itself ships as compiled binaries | Higgsfield's servers | Your plan credits; Unlimited does not apply | | Host your own Higgsfield-style studio | [app-templates](https://github.com/higgsfield-ai/app-templates) (inspect only until it has a licence) | None | UI on your server, generation on Higgsfield's | API balance | | Generate offline with no account | Open-weight models in a local tool such as ComfyUI; nothing from Higgsfield | Per model | Your GPU | Nothing per generation | The CLI and Higgsfield's hosted MCP server share your higgsfield.ai plan credits, while the API "does not use your higgsfield.ai plan credits" and runs on a separate prepaid balance ([Higgsfield blog](https://higgsfield.ai/blog/higgsfield-cli)). For agent setup, see our [Higgsfield MCP guide](/blog/higgsfield-mcp). If you want to ship a commercial studio today, build your own interface on the licensed SDKs. Higgsfield's studio template is worth reading for how it is put together, but it carries no licence, so reusing its code is not permitted until Higgsfield adds one. ## The audit: every public Higgsfield repository Higgsfield's verified GitHub organisation, [higgsfield-ai](https://github.com/higgsfield-ai) ("Higgsfield Inc.", linked to higgsfield.ai), had nine public repositories on 20 October 2026. | Repository | Licence | What it is | Generates on your machine? | What it needs | | ----------------------------------------------------------------------- | ----------------------------------------- | ------------------------------------------------------------------- | -------------------------- | ----------------------------------------------------------------------------- | | [higgsfield](https://github.com/higgsfield-ai/higgsfield) | Apache 2.0 | 2023 GPU orchestration and training framework for language models | Not a media generator | Your own GPU servers | | [cli](https://github.com/higgsfield-ai/cli) | MIT | Docs and an install script; the CLI ships as compiled binaries | No | Higgsfield sign-in, plan credits | | [skills](https://github.com/higgsfield-ai/skills) | MIT | Markdown skills for Claude Code, Codex and Cursor | No | The CLI and a Higgsfield account | | [cursor-plugin](https://github.com/higgsfield-ai/cursor-plugin) | MIT | Cursor marketplace entry pointing at Higgsfield's hosted MCP server | No | Higgsfield sign-in | | [higgsfield-client](https://github.com/higgsfield-ai/higgsfield-client) | Apache 2.0 | Python SDK for the API | No | API key and secret | | [higgsfield-js](https://github.com/higgsfield-ai/higgsfield-js) | No licence file; `package.json` says MIT | Node and TypeScript SDK for the API | No | API key and secret | | [app-templates](https://github.com/higgsfield-ai/app-templates) | None | Next.js studio template, installed through shadcn | No | API key from open.higgsfield.ai | | [omagotchi](https://github.com/higgsfield-ai/omagotchi) | MIT | Desktop avatar widget for the Omarchy Linux setup | No | Nothing for the default avatar; the CLI and plan credits to generate your own | | [homebrew-tap](https://github.com/higgsfield-ai/homebrew-tap) | No licence file; the formula declares MIT | Homebrew formula for the CLI | No | Same as the CLI | Read through the GitHub API on 20 October 2026, from each repository's licence file and package metadata. None of the nine runs image or video model inference on your machine; only the training framework downloads and runs model weights, and those are language models. On Hugging Face, the `higgsfield` account, which links to co-founder Yerzat Dulat's GitHub profile, holds language models and test uploads from October and November 2023 ([Hugging Face](https://huggingface.co/higgsfield)). Higgsfield's [`HiggsfieldAI`](https://huggingface.co/HiggsfieldAI) organisation there has no public models. Neither held image or video weights on 20 October 2026. **The "open-source" CLI's core code is not public.** Higgsfield's blog called the CLI "an open-source command-line client" on 19 October 2026 and said "the code and releases are on GitHub" ([Higgsfield blog](https://higgsfield.ai/blog/higgsfield-cli)). The [cli repository](https://github.com/higgsfield-ai/cli/tree/fd21e7cb550b5835cc00c6b4abfea0c02fe77ff2) holds an MIT licence, a README, a model list, a third-party notices file, a demo image and an [install script](https://github.com/higgsfield-ai/cli/blob/fd21e7cb550b5835cc00c6b4abfea0c02fe77ff2/install.sh). The program itself is published as compiled archives for macOS, Linux and Windows on each [release](https://github.com/higgsfield-ai/cli/releases) (v1.2.2 on 16 October 2026), and the `@higgsfield/cli` npm package is a small installer that downloads those archives. Its npm README lists the source as `github.com/higgsfield-ai/cli-src` ([npm](https://www.npmjs.com/package/@higgsfield/cli)), which returned 404 to signed-out requests on 20 October. The installer scripts are readable, and you may redistribute the binaries under MIT, but no source for the CLI program itself was in the public repository or the npm package on 20 October. **The studio template is public but unlicensed.** [app-templates](https://github.com/higgsfield-ai/app-templates), created on 24 September 2026, is a Next.js 16 studio "generating through the Higgsfield platform API", with one file per model and `HF_API_BASE_URL=https://api.higgsfield.ai` in its [example environment file](https://github.com/higgsfield-ai/app-templates/blob/2b6343a7f2cbe9f629aeb84e7fd98e7ad32a1af3/.env.example). It has no licence file, and its `package.json` is marked private with no licence field. Three users have opened issues asking for one ([#3](https://github.com/higgsfield-ai/app-templates/issues/3) on 7 October, [#5](https://github.com/higgsfield-ai/app-templates/issues/5) and [#6](https://github.com/higgsfield-ai/app-templates/issues/6) on 17 and 18 October), with no replies by 20 October. GitHub's terms let anyone view and fork a public repository, but without a licence, in GitHub's own words, "no one may reproduce, distribute, or create derivative works" from it ([GitHub Docs](https://docs.github.com/en/repositories/managing-your-repositorys-settings-and-features/customizing-your-repository/licensing-a-repository)). Read it freely; ask Higgsfield before you ship it. **The original Higgsfield project trains language models.** The [higgsfield](https://github.com/higgsfield-ai/higgsfield) repository, the organisation's most starred, is a GPU workload manager whose README example trains Llama 70B across your own nodes, with the model fully sharded by PyTorch FSDP ([llama.py](https://github.com/higgsfield-ai/higgsfield/blob/d12a36e66024a93d33ec61826a77d5a346c16869/higgsfield/llama/llama.py)). Alex Mashrabov announced it when he started the company in November 2023, writing "we open-source key frameworks that we build along the way" ([LinkedIn](https://www.linkedin.com/posts/amashrabov_github-higgsfield-aihiggsfield-fault-tolerant-activity-7130240128819601409-S6Vj)). Its last commit on the main branch is from 13 February 2024 ([commits](https://github.com/higgsfield-ai/higgsfield/commits/main)), and the newest PyPI release is a 0.0.4 release candidate from March 2024 ([PyPI](https://pypi.org/project/higgsfield/)). It is real Higgsfield open source, and it contains nothing from the image and video product. ## open-higgsfield, the repository announced as "everything" On 16 September 2026 Higgsfield launched a public developer API, ending its announcement with "Build your own Higgsfield with the best prices in GenAI industry" ([Higgsfield on X](https://x.com/higgsfield/status/2100305266688610746)). Minutes later @gpumaxxer, a personal account that X shows with Higgsfield's affiliation badge and whose bio reads "member of technical staff @higgsfield", quoted it ([@gpumaxxer](https://x.com/gpumaxxer/status/2100307886408912985)): ![An X post by @gpumaxxer, shown with a verified badge and a Higgsfield affiliation badge: 'Please build a Higgsfield competitor. We'll even power it. Higgsfield API is live. Cheapest on the market. Seedance 2.5 in the US, with face inputs. No other provider offers that. We also open sourced everything we've built over the last 18 months. The code behind a $5.4B startup is on GitHub. I'll personally give $50k to whoever builds something people actually fcking use. QT this with your demo. You have 7 days.' It quotes Higgsfield AI's 16 September post introducing the Higgsfield API.](https://cdn.moda.app/voyager/blog/is-higgsfield-open-source/fig-x-3e38a396b4d8f111.webp) The post behind the claim, 16 September 2026, captured signed out on 20 October 2026. The badge after the verified tick is X's affiliation badge for Higgsfield AI. The account's next post linked [wide-trace/open-higgsfield](https://x.com/gpumaxxer/status/2100307888959103232), and [a later one](https://x.com/gpumaxxer/status/2100334510470812157) added "FYI, this needs to be built with Higgsfield API". The following day Higgsfield's own account posted "Try building Higgsfield clone with our API" ([Higgsfield on X](https://x.com/higgsfield/status/2100629811131826198)). The claim came from the affiliated personal account, not Higgsfield's corporate account, and it described a release as already done rather than promising one. What the GitHub API showed for [wide-trace/open-higgsfield](https://github.com/wide-trace/open-higgsfield) on 20 October 2026, at commit [`b16a0ef`](https://github.com/wide-trace/open-higgsfield/tree/b16a0efe4d7e2707b56f8ccb02387fd2a9d2eddf): * **Owner:** `wide-trace`, a personal account with no profile details and no other public repositories. It is not the higgsfield-ai organisation, and nothing public says who operates it. * **Licence:** none. There is no licence file, GitHub's About sidebar shows none, and the README names none, though it calls the project "the free, open-source alternative to Higgsfield AI". * **What it runs:** a Next.js 16 web app with a prompt bar, per-model settings and a gallery. Each model file declares its own settings and input roles, the interface renders exactly those, and server actions map them to the API's fields; that catalog-driven design is the most reusable idea in it. * **Where your data goes:** the gallery history is kept in your browser's IndexedDB. Your API key goes to a server action ([actions.ts](https://github.com/wide-trace/open-higgsfield/blob/b16a0efe4d7e2707b56f8ccb02387fd2a9d2eddf/src/generation/actions.ts)) that stores it in an httpOnly cookie for 30 days ([credentials.ts](https://github.com/wide-trace/open-higgsfield/blob/b16a0efe4d7e2707b56f8ccb02387fd2a9d2eddf/src/generation/credentials.ts)), so whoever runs the server handles it on every request, and the server logs each request and response body, prompts and media URLs included ([platform.ts](https://github.com/wide-trace/open-higgsfield/blob/b16a0efe4d7e2707b56f8ccb02387fd2a9d2eddf/src/generation/platform.ts)). Reference images are uploaded to Vercel Blob with public access ([upload.ts](https://github.com/wide-trace/open-higgsfield/blob/b16a0efe4d7e2707b56f8ccb02387fd2a9d2eddf/src/generation/upload.ts)), so anyone with a file's URL can open it. On the hosted openhiggsfield.ai you are trusting an unidentified operator with both; the code shows the design, not what that deployment runs. * **What it calls:** generations go to whatever address you set in `HF_API_BASE_URL` (blank in the [example file](https://github.com/wide-trace/open-higgsfield/blob/b16a0efe4d7e2707b56f8ccb02387fd2a9d2eddf/.env.example)) with `Authorization: Key `. The README tells you to paste "your platform key as `id:secret`", the Higgsfield API's credential format, and its catalog of 38 models mirrors Higgsfield's: Soul, Seedance 2.5, Kling 3, MiniMax H3, Wan 3 and others. Pointing the base URL at a ComfyUI server would not work on its own: the client posts Higgsfield-style model requests and polls `/requests/{id}/status` ([platform.ts](https://github.com/wide-trace/open-higgsfield/blob/b16a0efe4d7e2707b56f8ccb02387fd2a9d2eddf/src/generation/platform.ts)), not ComfyUI workflows. * **Relationship to the official template:** every one of the 28 per-model files in Higgsfield's later app-templates ([catalog](https://github.com/higgsfield-ai/app-templates/tree/2b6343a7f2cbe9f629aeb84e7fd98e7ad32a1af3/generation/catalog/models)) has a same-named file [here](https://github.com/wide-trace/open-higgsfield/tree/b16a0efe4d7e2707b56f8ccb02387fd2a9d2eddf/src/generation/catalog), built on the same `ModelEntry` settings type, and both read `HF_API_BASE_URL`. That shows a shared design; it does not show who wrote which first. open-higgsfield has its own small server side, but none of Higgsfield's generation backend: no Cinema Studio, Genjutsu, Soul character training, credits system, moderation or model weights. "The studio itself is free", as its README says, in the sense that the web page costs nothing; pointed at Higgsfield, every image and clip is billed to your API balance. X readers have attached a context note to the September post saying the link is "not Higgsfield's proprietary code or backend", which matches what the repository shows. Higgsfield's help centre, on a page dated 5 September 2026, draws a line around its official surfaces: ![Higgsfield's help centre section 'Is Higgsfield open source?': 'No. The Higgsfield platform and its models are proprietary, and access is available only through the official surfaces listed above. Any code, app, or mirror site outside them that offers Higgsfield is not released or maintained by Higgsfield.' A second paragraph says Higgsfield is open in a different way, through public community projects, shared prompts and Higgsfield Originals assets on the website.](https://cdn.moda.app/voyager/blog/is-higgsfield-open-source/fig-help-fb2b870bc34ab62b.webp) Higgsfield's own answer, read on 20 October 2026. The wide-trace repository is not among the official surfaces the page lists. As for the lookalike domains from the opening: `openhiggsfield.ai` is linked from the wide-trace repository as its hosted version, and `thehiggsfield.ai` and `openhiggsfield.com` belong to a different project, listed below, that disclaims any Higgsfield affiliation. ## Comfy's challenge, and the projects that rebuild Higgsfield On 15 October 2026 Comfy Org, the company behind ComfyUI, launched a contest: "Higgsfield promised to open-source their entire platform. They didn't. Let's help them" ([ComfyUI on X](https://x.com/ComfyUI/status/2110870071933645103)). The [challenge page](https://comfy.org/open-source-higgsfield) offers about $25,000 in cash, GPUs and credits for open-source rebuilds of Genjutsu, Cinema Studio or another creative app built on the Comfy Developer Platform, requires entrants to "publish your app code with an open-source license and setup instructions", and closes on 30 October 2026 at 9am PT. Only its Open Weights Only category excludes partner nodes and closed APIs; other entries may use hosted models. The word "promised" is Comfy's: no first-party Higgsfield commitment to release Genjutsu, Cinema Studio or any model was found for this article, and no response to the challenge was on Higgsfield's [blog](https://higgsfield.ai/blog) or [changelog](https://higgsfield.ai/creator-hub/changelog) when checked on 20 October. Independent projects with Higgsfield in the name or pitch mostly follow one pattern: the interface runs on your computer, the generation does not. | Repository | Licence | Generation goes through | Affiliation statement | | ----------------------------------------------------------------------------------- | ------------------------------------- | -------------------------------------------------- | -------------------------------------------------- | | [SegFault42/HeliosGen](https://github.com/SegFault42/HeliosGen) | MIT stated in README; no licence file | kie.ai, or a Codex sign-in for GPT Image 2 | None found | | [BeatAPI/BeatDesign](https://github.com/BeatAPI/BeatDesign) | Apache 2.0 | BeatAPI | Calls itself "independent" | | [TechBeme/open-higgsfield](https://github.com/TechBeme/open-higgsfield) | MIT | Freepik, Gemini and other provider APIs | "Not affiliated with" Higgsfield | | [thehiggsfieldai/OpenHiggsfield](https://github.com/thehiggsfieldai/OpenHiggsfield) | MIT | fal or Higgsfield, with your own keys | "Not affiliated with" Higgsfield | | [synthetic-ape/comfield](https://github.com/synthetic-ape/comfield) | MIT | A ComfyUI server on your machine, or Comfy's cloud | Calls itself "Higgsfield-style"; a challenge entry | Read through the GitHub API and each README on 20 October 2026. "Local" in these READMEs usually means the interface and your files stay on your machine while prompts and reference images go to a provider: HeliosGen's feature list says "100% local" and "your data never leaves your machine", and on the next line, "Bring your own kie.ai key". comfield, created on 16 October, is the exception: its code [submits workflows](https://github.com/synthetic-ape/comfield/blob/3dbae26715869bc0c35507eedf840682f50e7c09/server/src/engines/workflow.ts) either to Comfy's cloud or to ComfyUI on your own machine at `127.0.0.1:8188`, and it [downloads open weights](https://github.com/synthetic-ape/comfield/blob/3dbae26715869bc0c35507eedf840682f50e7c09/server/src/local/downloads.ts) such as LTX and Z-Image into ComfyUI's folders for local mode. It was days old and was not run for this article. ## How to check any "open-source Higgsfield" repository Five checks, before you run anything: 1. **Affiliation.** Is it under `github.com/higgsfield-ai`, or linked from higgsfield.ai? Higgsfield's help centre tells you to treat anything you cannot find on its site as unofficial; a name or logo proves nothing. 2. **Licence.** Look for a licence file, a licence section in the README and a licence field in the package metadata. If none grants rights, you can view and fork the repository on GitHub but have no permission to reuse the code. 3. **Where generation runs.** Start with `.env.example` and the README, then follow the generate request into the code. A remote base URL means your prompts leave your machine; a local address such as `127.0.0.1:8188` points at a server on it. For what you will be charged, read the provider's own pricing and billing pages, not the repository. Then check what the tool stores: a provider API key is one thing, a copied browser session for your web account is another, and both give the operator the power to spend. 4. **Weights.** Local generation needs model files, bundled, downloaded by the app, or loaded by a local server you run. If the code shows none of the three, treat local generation as unproven. 5. **Where uploads go.** Reference images and faces sent to a blob store or provider leave your machine, whatever the README says about being local. Higgsfield's help centre warns that offers of "free Higgsfield" are "a common way to collect passwords and payment details" ([Higgsfield](https://higgsfield.ai/creator-hub/help-center/getting-started/official-higgsfield-platforms)). None of the projects above was found doing that, but the safe habit is the same everywhere: never give a third-party tool your Higgsfield password, and remember that an API key lets whoever holds it spend your balance. If you have already pasted a key into a site you do not trust, replace it in the official console at [open.higgsfield.ai](https://open.higgsfield.ai), which the help centre names as the place for "keys, the model catalog, usage, and billing". The GitHub CLI shows the owner, the detected licence and the environment file in two commands: ```text gh api repos/OWNER/REPO --jq '{owner: .owner.login, type: .owner.type, license: .license.spdx_id, created: .created_at, pushed: .pushed_at}' gh api repos/OWNER/REPO/contents/.env.example --jq .content | base64 -d ``` ## Why releasing interface code would not release the models Interface code does not include the models that produce the pixels, and many models in Higgsfield's catalog belong to other companies and are offered only as hosted APIs. Wan 3 is one example: as of 20 October 2026 the Wan-AI organisation on Hugging Face lists Wan 2.1 and 2.2 weights but none for Wan 2.5, 2.6, 2.7 or 3 ([Hugging Face](https://huggingface.co/Wan-AI); [Wan 3.0 in our model reference](https://voyager.so/resources/models/wan-3-0)). No weights for Higgsfield's own image and video models, such as Soul, were in its public repositories or Hugging Face accounts on that date. If what you want is offline generation, choose open-weight models rather than a Higgsfield repository. Wan 2.2's video models, for example, are downloadable under Apache 2.0 ([Wan2.2-T2V-A14B](https://huggingface.co/Wan-AI/Wan2.2-T2V-A14B)). Some models Higgsfield serves also have open weights, but read the licence: MiniMax H3's base weights are downloadable, yet its [community licence](https://huggingface.co/MiniMaxAI/MiniMax-H3/blob/main/LICENSE), dated 2 August 2026, excludes the European Union, the United Kingdom, South Korea and the United States from its territory. You run open weights in a local tool such as ComfyUI on a GPU with enough memory for the model, and you give up Higgsfield's presets, apps and closed models. For hosted products that compete with Higgsfield directly, see [Higgsfield alternatives](/blog/higgsfield-alternatives). ## Frequently asked questions **Is Higgsfield AI open source?** No. Its help centre says "there is no open-source version of Higgsfield AI" and that the platform and its models are proprietary ([Higgsfield](https://higgsfield.ai/creator-hub/help-center/getting-started/official-higgsfield-platforms)). As of 20 October 2026 its public repositories are SDKs, agent tooling, an unlicensed studio template and a 2023 training framework. **Did Higgsfield open-source its platform in September 2026?** No. A Higgsfield-affiliated personal X account [said](https://x.com/gpumaxxer/status/2100307886408912985) "we also open sourced everything we've built over the last 18 months" on 16 September and linked [wide-trace/open-higgsfield](https://github.com/wide-trace/open-higgsfield), an unlicensed interface built for Higgsfield's paid API, without model weights or Higgsfield's generation backend. **Can I self-host Higgsfield?** Only an interface. [app-templates](https://github.com/higgsfield-ai/app-templates) runs on your server but sends every generation to Higgsfield's API and bills your API balance, and it has no licence yet, so treat it as reference code. As of 20 October 2026 Higgsfield had published neither its platform backend nor its own models, and its help centre says they are available only through its official surfaces ([Higgsfield](https://higgsfield.ai/creator-hub/help-center/getting-started/official-higgsfield-platforms)). **Is the Higgsfield CLI open source?** It is MIT-licensed, but its [public repository](https://github.com/higgsfield-ai/cli) holds an install script, docs and compiled releases; the `cli-src` repository its [npm page](https://www.npmjs.com/package/@higgsfield/cli) names as the source was not public on 20 October 2026. **Is it safe to paste my Higgsfield API key into an open-source Higgsfield site?** Only if you trust whoever runs it: in open-higgsfield's code the key is held server-side and every request is logged ([platform.ts](https://github.com/wide-trace/open-higgsfield/blob/b16a0efe4d7e2707b56f8ccb02387fd2a9d2eddf/src/generation/platform.ts)). Running the code yourself avoids the third party; if a key may be exposed, replace it at [open.higgsfield.ai](https://open.higgsfield.ai). **What is Comfy's Open Source Higgsfield challenge?** A Comfy Org contest with $25,000 in prizes for open-source rebuilds of apps like Genjutsu and Cinema Studio on the Comfy Developer Platform, closing 30 October 2026 at 9am PT ([challenge page](https://comfy.org/open-source-higgsfield)). Higgsfield is not listed as an organiser or partner. ## Voyager is a desktop AI agent for creative work Describe what you want to make; Voyager plans it, uses your apps and tools, and keeps everything editable. [Sign up](/sign-up)