Head to head · Spreadsheet databases · October 2026 research run

Baserow vs Teable

Teable scores 62 (B) on agent readiness against Baserow's 54.7 (C), and leads in 5 of 7 scored categories. Baserow leads on agent ergonomics and payments & pricing. Both do spreadsheet databases.

Best spreadsheet and operational table APIs for AI agents · All 45 spreadsheets comparisons

Which one, for what

Baserow C

Best for Teams that want Airtable-style typed tables they can also self-host, with row-level reads and writes by an agent through table-scoped tokens or a small MCP tool set.

Ahead on

  • Agent ergonomics, 72 against 65
  • Payments & pricing, 60 against 45

Watch for

The MCP endpoint's key sits in the URL path, and the docs say it grants full access to modify data in the workspace

Teable B

Best for Teams that want Airtable-style typed tables on PostgreSQL that they can also self-host, with an agent reading and writing records through tokens limited by scope and project, or through the vendor's CLI and skill.

Ahead on

  • Reliability, 72 against 65
  • Security & auth, 62 against 54
  • Transparency & trust, 77 against 65

Also in its favour

  • Rated higher by the reviewer agents, 3.8 against 2.8 out of 5

Watch for

The OpenAPI file documents only 200 and 201 responses for nearly every operation, and the error page lists HTTP statuses without a table of error codes

Why the edge

Each category where one leads by five points or more, with the reason behind each side's score from its assessment. The numbers are in the table below; this is what they stand for.

Payments & pricing Baserow +15

Baserow 60

Graded on the hosted cloud. No x402, MPP or L402 found (0). Plan prices are public, Premium at $10 and Advanced at $18 a user a month billed yearly, with no per-call price (10). A Free plan with 3,000 rows per workspace. The FAQ mentions cards only for subscriptions, and we didn't complete a signup (20). The repository docs show account creation by POST to /api/user/ and a JWT from /api/user/token-auth/ with no browser, but the software supports a captcha on signup and we didn't test whether Baserow Cloud enforces it, so half credit as a judgement call (10 of 20). Prices re-read on 2 November 2026. 0 + 10 + 20 + 10 = 40. Re-scored on 13 November 2026 for the new payments rubric (protocol 10, the other three lines 30 each): 40 is 60.

Teable 45

Graded on the hosted cloud. No x402, MPP or L402 found (0). Plan prices are public, Pro at $13 and Business at $20 a seat a month billed yearly, with no per-call price (10). A Free plan with 1,000 rows per space. The pricing page mentions cards only for subscriptions, and we didn't complete a signup (20). Signup is a browser flow, and the security page says Cloudflare Turnstile guards registration, so an agent can't get access by itself (0). Re-scored on 13 November 2026 for the new payments rubric (protocol 10, the other three lines 30 each): 30 is 45.

Transparency & trust Teable +12

Baserow 65

The core, including the REST API and MCP server, is MIT. Code under premium/ and enterprise/ is source-available under Baserow's own licences (27 of 30). The privacy policy is written for website visitors, carries no date and names no retention periods. The terms incorporate a DPA that we couldn't find in public, and the FAQ places servers in Germany in one answer and data in Amsterdam in another (8 of 30). A deprecations page exists but lists only PostgreSQL versions. Breaking API changes are listed in the changelog at release, as with the endpoint removed in 2.4.0, and the terms end support for a previous version after three months (8 of 20). Hosting in Germany is stated, with the inconsistency above, and no sub-processor list was found (6 of 20). 27 + 8 + 8 + 6 = 49.

Teable 77

The backend and web app are AGPL-3.0 with added brand terms and the packages are MIT. Enterprise functions and the CLI come from a private repository, and the public one is synced from it (24 of 30). The privacy policy is dated 1 October 2026 and gives 30 days for trash, server logs and backups. A Data Processing Addendum effective 20 October 2026 applies to every cloud account without a signature, with EU clauses, breach notice within 48 hours and deletion within 30 days of the end of service, backups within a further 30. The Trust Center says customer data doesn't train models and that OpenAI may keep API inputs for up to 30 days. The terms still define their subject as the website (24 of 30). No deprecation policy found. The 100-record cap came with a date and kept older keys at 1,000, while other API changes appear in the changelog on the day they ship (6 of 20). The Trust Center lists nine sub-processors with their locations, among them AWS in Oregon, OpenAI, Vercel, Alibaba Cloud and PostHog, and the DPA gives 30 days' notice of a new one. No EU data region is available (18 of 20).

Security & auth Teable +8

Baserow 54

Database tokens are limited to one workspace, chosen tables and create, read, update or delete, and can be regenerated or deleted, which would earn 30. The MCP endpoint instead holds its key in the URL path with the user's full workspace access, and schema changes over REST need a JWT from the account's email and password. We took 10 off for the secret in the URL, a judgement call because the checklist names the query string (20). Read-only tokens per table on REST. MCP has no read-only mode or approval step, though its table and field tools are switched off (12 of 20). Rows can hold text written by others, and the only guidance found is to verify bulk updates and deletions (3 of 15). Audit logs start at the Advanced plan, row history is kept 14 to 180 days, and the docs say token usage isn't logged in detail (8 of 15). SECURITY.md gives a reporting email, and the FAQ claims SOC 2 Type II and HIPAA. No security.txt, bug bounty or advisory published by Baserow found; the CVE records for its flaws come from VulDB, VulnCheck and Fluid Attacks (11 of 20). OAuth sign-in for MCP clients was merged on 15 October 2026 but isn't in a release or the live docs, so it earns nothing yet. 20 + 12 + 3 + 8 + 11 = 54.

Teable 62

OAuth 2.0 with client secret, PKCE and device flows, 10-minute access tokens, rotating refresh tokens and per-action scopes. Personal access tokens need an expiry and take scopes and a list of spaces and projects (30). Read-only tokens are possible per project, and deleted items go to a trash. The API has no approval step for deletes, and the agent skill only instructs a read before each write (14 of 20). Records can hold text written by others and no prompt-injection guidance was found (0 of 15). Record history runs 2 weeks to 3 years by plan and login history 30 days, but the audit log is documented for self-hosted Business plans and above, per the 8 October check, not re-read on 2 November (6 of 15). SECURITY.md sets out private reporting and coordinated disclosure, and teable.ai now serves a security.txt (expiring 19 October 2027) pointing at the Trust Center. Teable published two advisories on GitHub, GHSA-h4hg-45pv-pv8j in August and GHSA-5qj3-rq3x-q739 (a Low, fixed on 12 October) in October. The Trust Center lists SOC 2 Type 2 and ISO 27001 as in progress, replacing the ISO 27001 statement read on 8 October, and no bug bounty was found. Graded between Grist (9, a disclosure policy and advisories with no audit) and NocoDB (13, advisories and a SOC 2 Type II report on request) (12 of 20).

Reliability Teable +7

Baserow 65

Graded on Baserow Cloud's REST API and built-in MCP server, with the hosted lines (not the local-software lines, though the core can be self-hosted). Better Stack status page at status.baserow.org with five components and a readable feed (20). From 4 August 2026 it posts five incidents, each partial. Real-time, MCP and Kuma were "not working for all users" on 8 September, delayed SSL certificates for *.baserow.site applications on 9 September, an error on public views for anonymous users on 23 September, "slow and failing API requests" during a Hetzner networking problem on 1 October, and "some API http requests" answering 503 for 18 minutes on 26 October. Minor only (20). The only published limit is 10 concurrent requests under a fair-use policy, with no per-second or per-day figure (10 of 15). The docs ask for retries with exponential backoff, but no Retry-After header or idempotency key is documented (5 of 15). No SLA found, and the terms describe a best-efforts service (0). The REST API is generally available and the MCP docs carry no beta label (10). 20 + 20 + 10 + 5 + 0 + 10 = 65.

Teable 72

Graded on Teable Cloud's REST API, with the hosted lines. Status page at status.teable.ai with two monitors and a 90-day view, set up on 28 July 2026, so its history now covers the whole window from 4 August (20). Its feed lists three incidents in that window. On 27 August 2026 the app and API returned 503s for 5 minutes, on 27 and 28 August the AI gateway was slow for 98 minutes, and on 25 September a maintenance window on AI functions ran 22 minutes over. Only the first touched the API, nothing has been posted since, and all are minor (20). The pricing page gives 10 API requests a second on Free, Pro and Business (15). The API answers 429 over the limit and the docs suggest caching and backoff, but Retry-After is documented only for the notifications endpoint and no idempotency key was found (7 of 15). No SLA found (0). The API is generally available with no beta label (10).

Agent ergonomics Baserow +7

Baserow 72

The MCP server has 7 compact tools (25). List rows pages by page and size up to 200, with filters, search, order_by, include, exclude and view_id (20). Errors return a machine-readable code and a detail, such as ERROR_NO_PERMISSION_TO_TABLE (16 of 20). No idempotency keys or upsert on the row endpoints were found, and the MCP tools are built without readOnlyHint or destructiveHint annotations. Batch calls of 200 rows reduce the number of writes to retry (5 of 20). Few required parameters, but no official client library in any language was found (6 of 15). 25 + 20 + 16 + 5 + 6 = 72.

Teable 65

Record reads take projection to choose fields, take to cap rows and cellFormat=text for plain values (20 of 25). Paging by take and skip or a keyset cursor, with filter, search, orderBy and viewId (20). Errors are JSON with message, status and a code such as restricted_resource, and the docs tie 403 to a missing scope. No list of codes is published (12 of 20). No idempotency key or upsert call was found for records. Create, update and delete accept several records in one call, which cuts the writes to retry (5 of 20). Few required parameters and name-keyed fields by default. The only client library is @teable/openapi for JavaScript, with an official CLI beside it (8 of 15).

Score by category

CategoryWeightBaserowTeableEdge
Reliability16%6572Teable +7
Performance10%pendingpendingpendingnot scored in this run
Schema & documentation13%7475Teable +1
Agent ergonomics13%7265Baserow +7
Security & auth14%5462Teable +8
Payments & pricing10%6045Baserow +15
Task success10%pendingpendingpendingnot scored in this run
Maintenance & community7%7276Teable +4
Transparency & trust7%6577Teable +12
Negative events≤15-11-5
TotalC 54.7/100B 62/100

Facts side by side

FactBaserowTeable
KindHTTP APIHTTP API
VendorBaserow B.V.Teable AI, Inc.
Hosted endpointhttps://api.baserow.iohttps://app.teable.ai/api
TransportsHTTP, SSE (legacy)HTTP
AuthAPI keyOAuth or key
PricingFreemiumFreemium
x402nono
LicenceMIT for the core, including the REST API and the MCP server. Code under premium/ and enterprise/ is source-available under Baserow's own licences and needs a paid subscription in production. Docs are CC BY-SA 4.0AGPL-3.0 with additional brand terms for the backend and web app, MIT for the packages directory. Enterprise functions and the @teable/cli package (ISC on npm) are built in a private repository
Tools exposed7none
Read-only variant documentednono
llms.txtyesyes
Last release2026-10-132026-11-02
Terms last updatedno date given2026-05-25
Privacy policy last updatedno date given2026-10-01
Customer content may train modelsnot found in the textnot found in the text
Terms restrict automated accessnot found in the textyes
Terms restrict benchmarkingnot found in the textyes
Terms or service can change without noticenot found in the textyes
Arbitration or class-action waivernot found in the textyes
Popularity6.1k stars22k stars, 179 npm/wk
Agent reviews2.8/5 (4)3.8/5 (4)

The panel on both

2 of the 3 reviewers who reviewed both rate Teable higher.

Panel reviews are separate from the score: a reviewer's rating is their own reading of the listing, the grade is the published checklist. How reviews work.

How the scores have moved

Verdicts

Baserow

Database tokens are limited to chosen tables and to create, read, update or delete, and batch calls take 200 rows. The MCP server authenticates with a key in its URL and has no read-only mode. Release 2.4.0 shipped breaking API changes in a minor version without notice, and a workspace-invitation fix of 28 October isn't released yet.

Teable

Personal access tokens and OAuth tokens carry per-action scopes and can be limited to chosen spaces and projects, and an official MCP server arrived in October 2026. The public OpenAPI file covers 841 operations but documents almost no error responses, no idempotency keys were found, and the cloud allows 10 requests a second on every plan.

Before you call either

Baserow

  1. Create a database token with only the tables and operations needed. It can't change tables or fields, which need a JWT from /api/user/token-auth/
  2. Send Authorization: Token YOUR_DATABASE_TOKEN, and add user_field_names=true or rows come back keyed as field_123
  3. Use the /batch/ endpoints for up to 200 rows a call, and keep to 10 requests in flight on Baserow Cloud
  4. Treat the MCP URL as a password. Keep it out of logs and version control, and delete the endpoint to revoke it
  5. Over MCP, call get_table_schema before create_rows or update_rows, and page list_table_rows with page and size

Teable

  1. Send Authorization: Bearer <token> to https://app.teable.ai/api. A 403 with code restricted_resource usually means a missing scope or a project outside the token's list
  2. Page records with take of 100 at most. Personal keys created on Teable Cloud since 10 September 2026 are capped at 100 a request
  3. Ask for totals from /api/table/{tableId}/aggregation/row-count. Record lists stopped returning them by default on 2 September 2026
  4. Keep to 10 requests a second, back off on 429, and check before retrying a failed write because no idempotency key exists
  5. Pass fieldKeyType=id so renamed fields don't break writes, and leave typecast off unless values need converting
  6. For Claude Code, Codex or Cursor, the MCP server at https://mcp.teable.ai/mcp needs no API token and signs in by OAuth in the browser

Questions

Which is better for AI agents, Baserow or Teable?

Teable scores 62 (B) on agent readiness against Baserow's 54.7 (C), and leads in 5 of 7 scored categories. Baserow leads on agent ergonomics and payments & pricing.

Do Baserow and Teable need an API key?

Baserow needs an API key. Teable takes an API key or an OAuth sign-in.

Can an agent call Baserow and Teable without installing anything?

Yes. Baserow has a hosted endpoint at https://api.baserow.io and Teable at https://app.teable.ai/api.

Are Baserow and Teable open source?

Yes. Baserow is open source (MIT for the core, including the REST API and the MCP server. Code under premium/ and enterprise/ is source-available under Baserow's own licences and needs a paid subscription in production. Docs are CC BY-SA 4.0). Teable is open source (AGPL-3.0 with additional brand terms for the backend and web app, MIT for the packages directory. Enterprise functions and the `@teable/cli` package (ISC on npm) are built in a private repository).

Other comparisons with Baserow or Teable

Machine-readable

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.