Cheap Seats vs Elmo

Cheap Seats (cheapseats.dev) is a personal open-source project: its author wrote it to make his own product better, drawing on lots of inspiration, Gauge first among them. It’s MIT licensed, free and always will be, with no hosted tier, and it runs on the Claude subscription you already have. Elmo is another open-source project in the same space: a self-hosted platform from Blue Whale Software, also MIT, and the open core of Elmo Cloud, from $29 a month. The two take different approaches. Elmo reaches the consumer surfaces, ChatGPT, Google AI Mode and AI Overviews, Gemini, Perplexity and Copilot, through scraping services, and checks each mention with TypeSafe’s Jev model. Cheap Seats asks only Claude and OpenAI models, through their APIs or the Claude Code and Codex CLIs, grades every answer recommended, mentioned or absent, puts an interval on every number, and ends each run with a ranked fix list. Elmo does more, and does several things better; this page says which.

Side by side

As of October 2026, from Cheap Seats’ docs and Elmo’s own README, providers, visibility, configuration and pricing pages:

What it costs. Cheap Seats is free and open source (MIT). Claude is required, through a subscription or an API key; ChatGPT (Codex or an OpenAI key) is optional and adds a second assistant’s answers. On your own machine, with a Claude Code subscription, answers, grading and the fix list can all run on the claude CLI, with no API key and no extra spend. Host it and you pay your host (and, in a container, API usage, since the image has no CLIs: about $0.03 to grade a 68-answer run, under a daily budget cap). Cost and budget has the details.

Self-hosted Elmo is free software too, but it calls paid services. Its providers page estimates a scraper at about $0.15 to $2.25 per prompt a month, at about 300 evaluations a prompt; direct model APIs at about $2 to $5 per tracked model per prompt a month; and Jev at about $0.15 to $0.35 per 1,000 answers, or $2 to $4 a month for 25 prompts on 3 platforms.

How the numbers differ

The two use some of the same words for different counts. Metric definitions has the full table; the short version:

• Visibility. Elmo’s is the share of each prompt’s latest answers that mention you. An answer saying it has no information about your brand isn’t counted. Cheap Seats’ named rate counts every graded answer that names you, and an answer that says it can’t find you still names you (graded mentioned, with the objection recorded).
• Recommended. Cheap Seats grades whether the answer endorses you, not just names you. Elmo reads sentiment toward you (positive, neutral, mixed, negative) but doesn’t grade recommendation.
• Share of voice. Elmo divides your mentions by yours plus your tracked competitors’. Cheap Seats divides by every product the answers named, so adding a competitor to your config doesn’t change your share.
• Position. Elmo averages where you rank among the tracked companies an answer names; Cheap Seats counts every product named.
• Sentiment. Elmo scores positive 100, neutral and mixed 50, negative 0, on a 0 to 100 scale. Cheap Seats’ net sentiment is the share positive minus the share negative, from −100 to +100.
• History. When you add a competitor or change a name, Elmo re-judges past answers for that company, 90 days back by default (REPROCESS_DAYS); older answers keep their old judgement, so a long chart can shift at that point. Cheap Seats keeps each run’s grades as graded, and marks a change of method on the trend instead.

Start with Cheap Seats if…

• You want to start at no cost: no scraper account, no TypeSafe key, nothing beyond the Claude subscription you have.
• You want each answer graded recommended versus merely mentioned, with the reasoning you can check, and what the models find on the web kept apart from what they already believe.
• You want an interval on every number and a test before a change is called a change, so a fluke doesn’t read as a trend.
• You want a fix list that lands as pull requests in your own repo, with each fix’s effect measured afterwards, plus free site checks.

When to move to Elmo

If you outgrow Cheap Seats, do whatever suits you. Its author would suggest Gauge; that team is great. If what you need is on this list, Elmo is a strong choice, and an open-source one:

• You need the consumer surfaces. Elmo scrapes ChatGPT as a person sees it, Google AI Mode and AI Overviews, Gemini, Perplexity and Copilot. Cheap Seats asks the models directly and covers none of the Google, Perplexity or Copilot surfaces. See Which AI engines Cheap Seats asks.
• You track several brands, or work for clients. Elmo is multi-brand, and its white-label option adds custom branding, a custom domain and SSO. Cheap Seats has one [brand] per config and no agency tier.
• You report to people who won’t log in. Elmo’s shareable reports open without an account.
• You want to build on it. Elmo has TypeScript and Python SDKs over its REST API, and a PostgreSQL database you can query directly.
• You want deeper query fan-out analysis. Cheap Seats’ searches page shows what the assistants searched, whether your sites came back, and which names they added. Elmo goes further across more engines: which words each one adds, drops or keeps, and which searches your content wins or misses.
• You’d rather not run anything, but still want the open-source code behind the numbers. Elmo Cloud runs the same platform for you.

Elmo is also well finished: a live demo, sentiment passages you can click through to, and published reliability checks of every scraper it supports.

Taking your prompts with you

Your prompt sets are plain TOML in your own git repo. Each prompt is a line of text with an id, so the buyer questions you’ve refined in Cheap Seats are yours. They’re plain text you can paste into Elmo’s prompt settings, or send through its create_prompts MCP tool or REST API, and your run history stays in your SQLite file.

See Getting started to run it, or all comparisons.