Plans & billing

tela meters usage per account and enforces per-tier limits. An account is either your personal account or an organization — each carries a plan, and quotas charge against whichever account owns a space.

What changes between tiers is how much AI you get (how many sources Atlas keeps fresh, how much indexing time a month, and how many answers), and — for organizations — seats and the Enterprise controls (SSO, audit log). Every plan ships the whole product: the wiki, Atlas, search, ask-your-docs, decks, public spaces, sync, and the agent connector. Pages and spaces are unlimited on every tier.

[!NOTE] Self-hosted? You don't pay for the software — see Self-hosting: editions & license below. The plan tables here cover telawiki.com (the hosted instance).

Plans

Personal

Organizations

Enterprise is unlimited usage plus the company-of-record controls (SSO, SCIM, audit, advanced roles and governance). Custom pricing — contact tela@telawiki.com.

Trial

New signups on telawiki.com get an automatic 30-day trial of Personal — no card required. During the trial the account runs with Personal features and Personal limits, including scheduled Atlas refresh and the 15-hour monthly indexing budget.

In the final week before expiry, plus a 1-week grace period after, an in-app banner warns you. When the grace runs out the account drops to Free automatically — nothing is deleted; you just can't add past Free's caps, and Atlas moves to manual refresh.

[!TIP] Convert from the trial to a paid plan at any time — you don't need to wait for it to expire. See [[How to upgrade]].

What's metered

Each limit charges the owning account of the space:

• Built-in AI answers / month — each ask your docs or chat answer generated by tela's own AI, per calendar month. Free includes a taste (50/mo); paid tiers raise it. This meters only the built-in assistant — asking through [[Agents & MCP|your own connected agent]] runs on your model and is not counted.
• Atlas indexing time / month — the limit most likely to affect you. Generating docs from a source is by far the most expensive thing tela does, so it's metered in the unit that actually costs: minutes of indexing. A single run ranges from about 5 minutes on a small repo to 80 on a large one, which is why a source count alone tells you nothing about what your schedule will cost.
  • Your refresh cadence multiplies it: a 50-minute repo refreshed daily is ~1,500 minutes a month, well past the Personal budget, while the same repo weekly is ~215. The Atlas screen projects your current schedule against your budget and, if it doesn't fit, tells you which cadence would.
  • A failed run costs you nothing — only runs that complete or are in progress count.
• Atlas sources & refresh — connected git repos or Jira projects Atlas keeps generated. On Free, sources refresh on demand — you click Refresh when you want an update. Paid tiers refresh automatically on a schedule you choose.
• Atlas run size and volume — alongside the time budget, each tier bounds files per run, runs per month, and embedding tokens per month.
  • The file cap is checked right after the repo is fetched, before anything is generated — an oversized source is refused in seconds and costs you nothing. Narrow it with the source's subpath or include/exclude filters and it will run.
  • Embedding tokens scale with how much text a source contains, not with its file count, and are spent mostly on a source's first full index; later refreshes cost far less. Watch the meter rather than estimating from file count.
• Semantic search & research — never monthly-capped, but rate-limited as fair-use per account (a generous per-minute budget) so the shared embedder stays responsive for everyone. Plain full-text [[Search & ask your docs|search]] is never capped at all. Editing pages re-indexes them in the background, and that is never counted.
• Seats — members of an organization (org plans only).
• Storage — total uploaded-file bytes across owned spaces. Generous on every tier; a backstop, not a headline limit.
• Pages & spaces — not limited. Create as many as you like on any tier.

An Unlimited limit applies to Enterprise. Settings → Plan & Usage shows each metric as used / cap with a progress bar, and the Atlas screen shows your indexing budget for the month plus what your current schedule is projected to use.

[!NOTE] Why hourly refresh isn't offered on metered plans. Hourly is about 730 runs a month. Even a very fast 2.5-minute run would spend more than 1,800 minutes — more than any metered plan includes — so the option would fail for every repo, not just large ones. Choose daily, weekly or monthly; Atlas only regenerates sources that actually changed, so a slower cadence rarely means staler docs. Hourly is available on Enterprise and on self-hosted instances, which aren't metered.

How limits behave

Quotas gate new growth only — you can always keep reading and editing what exists. Hitting a cap blocks adding more; the API returns quota_exceeded (HTTP 402). A fair-use rate limit instead returns rate_limited (HTTP 429) — wait a moment and retry. Existing content is never touched.

Budgets reset on the 1st of each calendar month. If you exhaust your indexing time, scheduled refreshes pause and resume automatically at the reset — your existing generated pages stay exactly as they are, and you can still run a source manually once the budget returns.

Self-hosting: editions & license

Self-hosting is open core — you don't pay for the software:

• Community (free, AGPL) — the whole product: wiki, Atlas, ask-your-docs, semantic search, decks, public spaces, custom domains, sync, agents/MCP, organizations and per-space roles. AI is bring-your-own — point tela at your own LLM and embedder; nothing is metered (and scheduled Atlas refresh, including hourly, runs for everyone). An instance admin can assign any tier from Settings → Plan & Usage or the set-plan CLI.
• Enterprise — adds the company-of-record layer (SSO / SAML / OIDC, SCIM, audit log, advanced roles and premium connectors) via an offline license key, $8 per seat / month, billed annually (the price of a Team seat — you run the infrastructure and bring your own AI). Without a valid key these features stay locked. Seats are honor-system — over your count, the admin sees a notice, never a lockout.
• Commercial license — relief from the AGPL's copyleft for companies that need it, with no change to features.

Buy a self-host Enterprise license — self-serve. On telawiki.com → Settings → Self-host licenses, choose your seats and check out. Your license key is emailed to you and shown in that tab; paste it into your own instance at Settings → License (instance admin), or set the TELA_LICENSE_KEY environment variable.

[!TIP] Renewals install themselves. If your instance can reach telawiki.com, it auto-pulls the renewed key each term — no manual re-paste. Air-gapped? Re-install the emailed key before the current one expires (there's a 14-day grace).

Contact tela@telawiki.com for a commercial license or volume pricing.

Upgrading and cancelling

• [[How to upgrade]] — step-by-step checkout for personal and org accounts
• [[Cancellation & downgrade]] — access timeline, what's preserved, and how to move to a lower tier