Billing & Pricing

How billing works

syndicAI uses a prepaid credit system combined with weekly GPU budgets. There are no subscriptions — you top up credit, pick a tier that caps your weekly GPU usage, and only pay for the hours your server actually runs.

  1. Add credit to your account via Stripe (one-time or auto top-up)
  2. Pick a tier that sets your weekly GPU hour budget
  3. Use your server — credits are deducted every 15 minutes based on actual GPU time
  4. Budget resets weekly on Monday at 00:00 UTC

Why GPU-hour billing is cheaper

Each syndicAI squad server runs on a dedicated 8× RTX 5090 setup (254.7 GB VRAM), running MiniMax M2.5 AWQ at well above 30 tokens/sec. In one hour, the server can process hundreds of thousands of tokens. In a 4-hour session, that's easily over 1 billion tokens of throughput capacity.

With per-token pricing, 1 billion tokens through Claude Sonnet would cost roughly $2,640. With GPU-hour billing through syndicAI at ~$3.28/hr, a 4-hour session costs ~$13. The savings are 5–70× depending on the API you're comparing against.

syndicAI sources GPU time from the spot market at competitive rates — we handle provisioning, deployment, security, and lifecycle so your squad gets dedicated hardware without the operational burden.

Weekly GPU budget tiers

Tier Weekly budget Daily average Max monthly cost Per user (5-person squad)
Starter 10h/week ~2h/day ~$142 ~$28
Standard 20h/week ~4h/day ~$284 ~$57
Pro 40h/week ~8h/day ~$568 ~$114

The tier defines a weekly cap — not a fixed charge. If your squad only uses 12 of 20 hours in a week, you only pay for 12 hours. The budget prevents runaway spending.

All tiers include the same features:

  • Frontier-class open-source models (your choice)
  • OpenAI-compatible API endpoint
  • API key management for each squad member
  • Usage dashboards with real-time credit tracking
  • Auto-start and auto-stop
  • Managed security with TLS
  • Squad cost splitting

Budget exhaustion

When your squad's cumulative GPU hours for the week reach the budget:

  1. The server auto-stops (in-flight requests are completed)
  2. The dashboard shows a "budget exhausted" status
  3. The server cannot be restarted until Monday 00:00 UTC when the budget resets
  4. No overage charges — the budget is a hard cap

If you consistently hit the limit, upgrade your tier for more weekly hours.

Near-budget warning

When 80% of your weekly GPU budget is consumed, the dashboard displays a warning so you can pace your remaining usage or upgrade proactively.

Credit management

Adding credit

Top up your account via Stripe Checkout. You choose the amount — there's no minimum top-up beyond the initial $10 minimum balance required to create a Squad Server.

Auto top-up

Enable auto top-up to avoid interruptions:

  • Set a threshold (e.g. $10) — when your balance drops below this, a charge is triggered
  • Set a top-up amount (e.g. $50) — the amount charged to your saved payment method
  • Requires a saved payment method via Stripe

Zero balance

If your credit balance reaches zero while a server is running:

  1. All running instances are stopped immediately
  2. Syndicates enter "credit exhausted" status
  3. Top up your credit to restore access and restart manually

Credit ledger

Every financial event is recorded in an immutable credit ledger — top-ups, usage deductions, refunds. View your full history via the dashboard or GET /api/credits/ledger.

Squad cost splitting

syndicAI supports two cost splitting modes:

Equal split

Squad members share the cost equally. The owner's credit balance is used, and members reimburse their share.

Example: Standard tier maxing out at ~$284/month with 5 members ≈ ~$57/member/month.

Owner-pays

The squad owner's credit balance covers the full cost. Other members use the server at no charge.

Coming soon: Usage-based split

Costs split proportionally based on each member's token usage. Stay tuned.

Refunds

  • Unused credit: Credit balances are non-expiring and can be used at any time
  • No partial-hour refunds: GPU time consumed is non-refundable
  • Disputes: Contact support for billing disputes — we handle these on a case-by-case basis