The Best Open-Source AEO Tools (2026)

The most complete open-source AEO tool is Elmo: MIT-licensed, free to self-host, and auditable down to how each metric is computed. Smaller open-source projects like OneGlanse, GEO/AEO Tracker, and Gego exist too, and one entry on this list — GetCito — is now simply a copy of Elmo's codebase with the copyright notice changed. Open source matters here because you can read the code, own your data, and avoid a black-box visibility score.

This is a developer's list. It covers answer engine optimization tools whose source is public and that you run yourself, not free tiers of closed products. For those, see our guide to free AI visibility tools. For the wider field of paid and hosted options, see the best AEO tools. And for the live directory of projects we track, see open-source AI visibility tools.

The honest picture is that this is a thin, early space: one mature platform, a handful of small projects, and the option to script your own checks. Below is what actually exists in 2026.

Key takeaways

Most AEO tools are closed and hosted. You send them your prompts, and you trust the number they hand back. Open source changes both halves of that. You can read exactly how a visibility score is built, and you can run the whole thing on hardware you control, so your prompts and history never leave your environment.

For a number that might land in a board report or set a content budget, that auditability is the point. Nobody should have to take a black-box metric on faith. Owning the data matters just as much: no vendor holds your visibility history, and there is nothing to migrate off if you decide to leave. The trade is upkeep. You run the infrastructure, and you keep it current as engines change.

Details and licenses reflect public information in mid-2026. Confirm current specifics in each project's repo.

The section above is our read. This one is not — it is what the GitHub API returns, so you can check every cell yourself. Figures pulled August 11, 2026, sorted by stars.

Commit counts are totals on the default branch and include automated commits; Elmo's 862 includes 67 from Dependabot and GitHub Actions, and GetCito's 35 includes 8. Contributor counts exclude bot accounts entirely. "Beyond the lead dev" is contributors minus the single largest committer — the cheapest available test of whether a project is a team or one person.

Elmo leads on every durable measure: most stars, most contributors, most contributors beyond one person, most releases, and the longest continuous history. Nine people have landed code in Elmo besides the lead developer. Across all five other projects combined, that number is four, and one of those four is an AI agent account.

Two numbers in that table need context, and both cut against a simple reading.

Canonry's 500 commits in 90 days is the highest on the list, and it is one person. 451 of those 500 came from the same developer committing under two names, and another 45 from an in-house AI agent account. It is also the one project here that is not actually open source — FSL is source-available until it converts to Apache 2.0. Real momentum, but a single-vendor project half Elmo's age.

OneGlanse's 496 commits landed almost entirely in one burst, between April 15 and May 10, 2026. It has not had a commit since, and it has never cut a release. High commit counts and sustained maintenance are different things.

Then there is GetCito, where the numbers tell the story better than any argument. Its repository contains 35 commits, 27 of them from humans, across its entire lifetime — for a codebase of well over 100,000 lines. That is because 131,716 of those lines arrived in a single import of Elmo's code in July 2026. Stars measure attention, not work.

Elmo is the most complete open-source AI visibility platform, and the reason this list has a clear top pick. It is released under the MIT license, free to self-host, and every metric is computed by code you can read. It tracks how AI answer engines mention and cite your brand across ChatGPT, Claude, Perplexity, Gemini, and Google's AI Overviews, among other engines, then turns that into a visibility score, citation analytics, brand-mention tracking, and competitor benchmarking. You can export everything through the API, and agencies can white-label it.

It fits teams that want to own their AEO data outright. The self-hosted core runs on Docker and PostgreSQL. You supply your own model API keys, which carry usage costs, and there is no license fee or per-seat charge. Be clear on what it is not: Elmo does not do sentiment analysis, prompt-volume estimates, content generation, shopping-result tracking, or geographic breakdowns. It measures visibility and citations well, and leaves the content work to you. A managed cloud option is listed as coming soon, for teams that would rather not run the infrastructure.

OneGlanse is an MIT-licensed, self-hosted tracker covering ChatGPT, Gemini, Perplexity, Claude, and Google AI Overview. Its distinguishing choice is how it collects answers. Instead of hitting the model APIs, it captures responses through authenticated accounts on the AI web interfaces, which is closer to what a real user sees. Data lands in a ClickHouse backend on your own infrastructure, and you bring your own API keys.

It suits developers who care that the answers being scored come from the actual chat products, not the API surface, and who are comfortable standing up ClickHouse. The idea is good and the approach is genuinely different from everything else here.

Check the activity before you build on it, though. Its 496 commits landed in a single burst between April 15 and May 10, 2026, and the repository has not had a commit since. It has one contributor and no tagged releases.

GEO/AEO Tracker is an MIT-licensed, local-first dashboard for watching brand visibility across six platforms: ChatGPT, Perplexity, Gemini, Copilot, Google AI Overview, and Grok. It stores everything client-side in IndexedDB, with no external database to run, and you provide your own keys for data fetching and model inference.

The local-first design is the appeal. Nothing leaves your browser, setup is light, and the platform coverage is broad for a solo project. The flip side is scope. It is a single-user tool built around one person's dashboard, not a team platform, so treat it as a personal monitor rather than shared infrastructure.

Canonry is the most ambitious of the independent projects. It is a self-hosted, agent-first AEO platform that tracks how Gemini, ChatGPT, Claude, Perplexity, and local models cite your site, ingests server logs to measure AI-driven traffic, and integrates with Google Search Console, GA4, Bing Webmaster, and Google Business Profile. A built-in agent named Aero exposes a 67-tool MCP adapter, and clients are configured declaratively in YAML.

One honest caveat on the "open source" label. Canonry ships under the FSL-1.1-ALv2, a source-available license that converts to Apache 2.0 after two years. You can read and self-host the code today, but it is not OSI open source in the strict sense until that conversion. If server-log attribution and agent workflows matter to you, it is worth a look, with that license nuance understood.

If you are considering GetCito, use Elmo instead. It is not its own project. On July 10, 2026, GetCito replaced its codebase with Elmo's in one pull request that changed 847 files and was self-merged 62 seconds after opening. Its LICENSE.md is Elmo's, with "Copyright (c) 2026 Blue Whale Software, LLC" changed to "Copyright (c) 2026 GetCito" — retaining that notice is the one condition MIT attaches to a fork.

The copy is still legible in the repo. GetCito's AGENTS.md opens "Elmo is an open-source AI visibility tracking platform," its CLA names Elmo's parent company, and its CODEOWNERS assigns every path to Elmo's founder. The AI Crawlability Clinic it used to be known for was part of the Next.js application that got deleted.

There is no reason to run a rebranded fork of a project you can run directly, especially one that is falling behind it: 17 commits to Elmo's 98 in the month after the copy. Full evidence in GetCito vs Elmo.

Gego is a GPL-3.0 GEO tracker you self-host from Docker Hub. It schedules prompts across generative engines on a cron, then captures citations, brand mentions, and extracted keywords, with competitor and alias tracking alongside them. It queries OpenAI, Anthropic, Google, and Perplexity, and it will also point at Ollama or a custom provider, so you can run the whole loop against local models. There is a dashboard, a CLI, and a REST API.

Two caveats. A full deployment expects PostgreSQL, MongoDB, and etcd, which is a heavier footprint than a single-container deploy. And GPL-3.0 is copyleft rather than permissive, so unlike the MIT projects here it constrains what you can build on top of it and redistribute. A managed Gego Cloud is on an early-access waitlist.

The real choice is not open source versus paid. It is who does the work and who holds the data. Enterprise AEO software like Profound gives you a hosted dashboard your team can open tomorrow, managed engine coverage, and a vendor who keeps it running. You pay a subscription, usually metered by prompt or seat, and the scoring is typically a black box you cannot inspect. Your visibility history lives in their system.

An open-source tool you self-host inverts that. There is no license fee, only your infrastructure and the AI provider API usage. You can audit every metric, and your prompts and history stay on your own machines. The cost moves from a subscription line to engineering time. You deploy it, you maintain it, and you own the coverage as engines change.

For a regulated team, a data-sensitive brand, or an agency that wants to white-label the whole thing, transparency and ownership usually win. For a team that needs answers this afternoon and has budget instead of engineering hours, a managed platform is the pragmatic call. Open source is not a black box, and that is precisely its trade: you see everything, and you run everything.

Start with how you will use it, not a feature grid. If you want full control, auditable numbers, and no per-seat fees, a self-hosted open-source tool fits, and Elmo is the most complete one on offer — and the one the public GitHub numbers actually favor, on stars, contributors, releases, and sustained history alike. If you want the same open code with the smallest footprint, GEO/AEO Tracker's local-first design is worth a look, bearing in mind it is one developer and 28 commits. If your priority is server-log attribution, Canonry goes furthest, with its license caveat and single-maintainer concentration in mind.

The one thing worth weighing above features is whether anyone will still be maintaining the project in a year. That is what the contributor column is for. A tool with one committer is one person's spare time away from being abandoned, and in a field where models get renamed and APIs shift every few months, an unmaintained tracker quietly stops being correct rather than visibly breaking.

Then weigh coverage, maintenance, and your own capacity to run infrastructure against what a managed subscription would cost. If the honest answer is that you have no time to self-host, that is useful to know early: see the best AEO tools for hosted options, or the free AI visibility tools guide for what a zero-budget setup can and cannot do. Whatever you pick, the job is the same. Get a reliable, repeatable read on whether AI answers cite you, from a tool you trust because you can see how it works.