Billing & tokens

Tablize has two billing concepts that work together:

• Plan subscription — unlocks features, storage quotas, integration access.
• Token allowance — how much the Agent can think on your behalf. Rolling windows on every plan.

Plus a third, optional:

• Top-up packs — one-time token purchases that don’t expire and serve as a bypass when a window is saturated.

Plans at a glance

	Free	Plus	Pro	Max
Price	$0	$20/mo	$60/mo	$200/mo
Persistence	Ephemeral (7 days)	Persistent	Persistent	Persistent
Connect DBs / APIs	—	✓	✓	✓
Watches	—	—	✓	✓
IoT / MQTT	—	—	✓	✓
Apps	—	✓ (3)	✓ (25)	✓ (unlimited)
Public sharing	—	—	✓	✓
Federation	—	—	—	✓
Priority compute	—	—	—	✓

See the Pricing page for the full matrix with storage limits and seat inclusions.

Token allowances

Each plan includes a rolling token allowance. Tokens are consumed by Agent inference — every SQL query, Python call, fetch, generation, etc. has a token cost (roughly proportional to time × complexity).

Tokens replenish on two overlapping windows:

• 5-hour window — a short burst allowance, mostly for interactive chat.
• 7-day window — a larger steady allowance, for scheduled Reports, Watches, and Scripts.

Both windows roll continuously — there’s no “reset day.” Use more in a given 5-hour stretch, that 5-hour window tightens; usage in the 7-day window is tracked independently.

Why two windows? Because a burst of activity (exploring data for an afternoon) is different from sustained activity (scheduled reports every hour). Separate windows let each kind of usage stay smooth.

What costs tokens

• Agent inference — every turn. Rough cost by turn type:
  • A simple lookup (SELECT + explanation): low.
  • A complex analysis (multi-step SQL + Python): medium.
  • App generation: high (the Agent writes HTML, CSS, JS).
  • Image generation: high.
  • Vision extraction from Media: high per image.
• Scheduled runs — Scripts, Watches, and Reports that re-run cost tokens each time they run. Scheduling a report every hour is ~24x the cost of running it once a day.
• Integrations — sync is free on Pro and Max. On Plus, large historical backfills may dip into tokens.

What doesn’t cost tokens:

• CRUD on Apps — direct PG, no Agent in the path.
• Reading cached query results — the Agent caches per-session; re-querying the same data is ~free.
• Looking at existing Reports / Dashboards — panels serve from cache. Scheduled runs regenerate.
• Viewer activity — Viewers don’t use tokens at all.

The token wallet

Sidebar → Billing → Usage. Shows:

• Current 5-hour window usage + remaining.
• Current 7-day window usage + remaining.
• Last 30 days of token usage, by type (interactive / scheduled / generation).
• Any top-up packs on file.

When a window tightens to < 20% remaining, the in-app banner warns you — once per window, not per session.

Top-up packs

Top-ups are optional one-time purchases:

• $10 pack — ~500K tokens.
• $50 pack — ~3M tokens.
• $200 pack — ~15M tokens.

They never expire. When your rolling window saturates, top-up tokens cover the overflow automatically. When the window refills, top-ups are paused.

Top-ups are the right tool for occasional spikes (a big client project, a one-off data migration). They’re not the right tool for steady-state usage — if you consistently overflow the window, upgrade the plan.

Viewer seats

$5/month per seat. Billed monthly alongside your plan. No token consumption. See Workspaces & roles.

Free plan, specifically

The Free plan is the full agent, not a demo. What you get:

• Enough tokens to finish one real question with 3-5 follow-ups.
• CSV / XLSX / JSON upload up to 50 MB.
• Shareable link to the answer.
• The Agent, same capabilities as paid plans (no feature gating).

What you don’t get:

• Persistence. Your workspace is ephemeral — data, sessions, and assets are deleted after 7 days of inactivity.
• Database connections. Upload files only.
• Integrations.
• Scheduling. Nothing re-runs itself.

The Free plan is designed to let you answer one good question with your actual data before deciding whether to upgrade. It’s not a long-term free tier.

Anti-abuse

Email normalization and browser fingerprinting deduplicate Free workspaces — same user creating multiple Free accounts won’t get infinite tokens. Token consumption records are retained indefinitely, even after a Free workspace is deleted.

Upgrades and downgrades

• Upgrade — takes effect immediately. Prorated — you pay the difference for the remainder of the current period.
• Downgrade — takes effect at the end of the current period. You keep current features until then, then lose access to anything the new plan doesn’t include.
• Cancel — takes effect at the end of the current period. Workspace becomes read-only after that; data retained for 30 days before deletion.

On downgrade, if you have more assets than the new plan allows (e.g., 50 Apps on Pro → downgrading to Plus’s 3-App limit), the UI walks you through archiving the extras.

Invoices and payment

• Stripe for all billing.
• Credit card, bank transfer, or SEPA supported.
• Invoices emailed to Owner monthly. Downloadable from Settings → Billing.
• VAT / tax collected for applicable regions, shown on invoice.
• Corporate billing info (company name, address, VAT ID) configurable.

Self-hosted billing

Self-hosted Tablize is a one-time purchase (or perpetual license) plus optional upgrade maintenance. Different model entirely — no token metering, because the LLM calls go through your API key. See Self-hosted.

Common gotchas

• “Out of tokens” mid-conversation. The 5-hour window saturated. Either top up or wait for the window to roll. Top-up is instant; the rolling wait is at most a few hours.
• Scheduled Reports suddenly stopped running. Usually a lapsed card on Stripe. Settings → Billing shows the status; update the payment method.
• Token usage looks higher than expected. Open the Usage drawer — it breaks down by category. Most common culprits: too-frequent Watches, App generation loops during experimentation, or image extraction at scale.
• Got a “overage” charge I didn’t authorize. Impossible on Cloud plans — they’re not metered. Check the invoice line items; if it’s genuine, it’s either a top-up pack you bought or a plan upgrade that prorated.

Next steps

• Pricing page — for the full plan matrix.
• Federation — unlocked on Max; adds cross-workspace compute.
• Self-hosted — avoid token metering entirely by running on your own box.