Comparison
Flitch vs Metabase
An AI-first dashboard builder against the best-known open-source BI tool. Where free self-hosting wins, and where it does not.
Which should you choose?
Flitch generates a web application from a description, with AI included on every plan and a semantic layer behind it, and publishes it on your own domain. Metabase assembles saved questions into dashboards and charges separately for AI. Metabase is the better choice if you can self-host, where the open-source edition is free for unlimited users, or if your team wants to work in SQL directly.
Checked against both products on .
Flitch vs Metabase at a glance
| Flitch | Metabase | |
|---|---|---|
| Entry price | Free tier, then $20/month flat (Pro) | Free self-hosted, or $90/month + $6/user (Starter, 5 users included) |
| Self-hosting | No | Yes, free, unlimited users |
| Terms | Proprietary SaaS | AGPL v3 for the open-source edition |
| Higher tiers | $30/user/month (Team) | Pro $517.50/month + $12/user; Enterprise from $20,000/year |
| AI included | Core to the product. The plan fee is the credit allowance: $20 buys 200 credits worth $20 | A separately priced service, $3.75 per 1m tokens |
| How dashboards get made | Described in plain language, generated as editable code | Built from questions and SQL, assembled into dashboards |
| How the numbers are computed | Generated SQL, aggregated at the source against the semantic layer | SQL you write, or the query builder, saved as questions |
| SQL | Written for you at generation, against the semantic layer. Custom SQL supported | First-class, the core of the product |
| Data sources | 61, including SaaS apps, file upload, an AI API finder and input tables | 18 official connectors, all databases and warehouses, no SaaS |
| Public publishing | Custom domains, password protection, email-restricted sharing | Public links and iframe embedding |
| Governed models | Semantic layer with relationships and calculations | Models, metrics and segments |
| Enterprise identity | No SSO or SCIM | SAML on Pro and Enterprise, not on Starter |
| Costs beyond the sticker price | None. Credits are the whole bill | AI at $3.75/1m tokens, storage from $2 per 1m rows, transforms $0.01 a run past 1,000. Self-hosting adds your infrastructure |
| Write-back | Input tables, dashboards that accept data entry | Actions on some databases |
Is Metabase really free?
Yes, if you self-host. The open-source edition is free under AGPL v3 for unlimited users, so if you have an engineer who can run a container your software bill is zero indefinitely. That is a genuine advantage and worth stating plainly. The costs are indirect. Someone deploys it, upgrades it, backs it up and keeps it available, and AGPL v3 obliges you to share modifications with users of your deployment, which matters if you embed a modified version commercially. Flitch also has a free tier, at 40 credits a month with 100 dashboards and 10 integrations, with nothing to run. So the choice is not free against paid; it is free-if-you-operate-it against free-with-no-infrastructure.
What does Metabase Cloud cost compared with Flitch?
Metabase Cloud Starter is $90 per month plus $6 per user with five users included, so a five-person team starts at $90 and a ten-person team is around $120. Pro jumps sharply to $517.50 per month plus $12 per user with ten included, and Enterprise starts at $20,000 per year. Flitch is $20 per month flat on Pro, or $30 per user per month on Team, so a five-person Flitch team is $150 against Metabase Starter at $90. For a single person or a pair, Flitch Pro at $20 flat is the cheaper hosted option. Past about three users, Metabase Cloud Starter is cheaper, and it stays cheaper until you need something on the Pro tier.
How differently do the two actually work?
Metabase is question-first. You ask a question of a table, refine it in the query builder or in SQL, save it, and assemble saved questions into a dashboard. Analysts who like SQL like Metabase a great deal, and it is excellent at that. Flitch is description-first, and it writes the SQL for you. Generation goes through a semantic layer holding your entities, relationships and calculations, so the queries behind a dashboard resolve against definitions rather than being improvised each time, which is what makes the output trustworthy without anyone reviewing every query. You can still write custom SQL when you want it. The other difference is what you end up with. Metabase gives you charts arranged inside Metabase. Flitch gives you a web application as code, on your own domain, which can take data in through input tables as well as display it.
Which handles non-database sources better?
Flitch, and the gap is wider than the headline counts suggest. Metabase lists 18 official connectors and every one of them is a database or a warehouse, so it expects your data to already be in one. On that ground the two are close: Flitch reaches 18 databases and 3 warehouses, with different specifics, since Metabase covers Oracle, ClickHouse, Presto and Athena while Flitch covers managed Postgres and MySQL variants Metabase does not list separately. The difference is everything outside that ground. Flitch has 61 integrations including SaaS applications directly, so Stripe, Xero, HubSpot, Salesforce, Intercom and similar tools connect without you first building a pipeline into a warehouse. Beyond connectors it also takes a file upload, will find and pull a public API for you, and offers input tables where a dashboard collects data from the people using it. If your data already lives in Postgres or Snowflake, none of this matters. If it lives in your accounting and CRM tools, or in a spreadsheet, or does not exist in a system yet, it matters a great deal.
Which is better for sharing dashboards outside your team?
Flitch, though Metabase is not weak here. Metabase offers public links and iframe embedding, which covers most internal and some external cases. Flitch publishes to a URL you can put on your own custom domain, with optional password protection or access restricted to named email addresses, and viewers can question the data with Ask AI rather than only reading it. If the dashboard is going to a client, or onto a public site as something close to a product, that difference is the reason to pick Flitch.
Choose Metabase if
- You can self-host, in which case it is free for unlimited users
- Your team wants to write and refine SQL directly, as the main way of working
- You want SAML, on the Pro or Enterprise tier
- You need a connector Flitch does not have, such as Oracle, ClickHouse, Presto or Athena
Choose Flitch if
- You have nobody to run and maintain infrastructure
- You want the dashboard generated rather than assembled question by question
- Your data is in SaaS tools and you have no pipeline into a warehouse
- The dashboard is going to a client or onto your own domain
- You are one or two people, where $20 flat beats $90 plus per-user
- You want AI included rather than metered as a separate service
Flitch vs Metabase FAQ
Is Metabase free forever?
The open-source edition is, under AGPL v3, for unlimited users. You pay in infrastructure and maintenance rather than software fees. Metabase Cloud is the paid hosted option, starting at $90 per month plus $6 per user.
Does Metabase have AI features?
Yes, as a separately priced service billed at $3.75 per million tokens with one million included. It is a capable addition to a SQL-first tool rather than the core of the product, which is the main design difference from Flitch.
Does Flitch use SQL?
Yes. Flitch writes the SQL when it generates a dashboard, resolving it against a semantic layer that holds your entities, relationships and calculations, so the queries follow definitions rather than being improvised. Custom SQL is supported too. The difference is that Metabase is built for people who want to write and refine queries by hand, and Flitch is built to remove that step.
Do the AGPL terms matter?
Only if you modify Metabase and expose it to users, typically when embedding a modified version in a commercial product. AGPL v3 obliges you to make those modifications available. For ordinary internal use it has no practical effect.
Which connects to more data sources?
Flitch, at 61 against 18. On databases alone they are close, since Metabase’s 18 official connectors are all databases and warehouses and Flitch reaches 18 databases plus 3 warehouses. The gap is everything else: Metabase has no SaaS connectors at all, while Flitch adds those plus file upload, an AI API finder and input tables.