Pricing
BatchRouter shows provider base rates on Models, then adds a clear service fee when a customer accepts a batch quote. The fee is disclosed on the quote and the batch receipt, not folded into the public model showcase.
Fee percentages
These are the current customer-facing service fee percentages. The quote preview still shows the estimated dollar fee before submission.
Applied to regular model-routing quotes as a percentage of provider subtotal.
Applied when a packaged workflow adds shaping, validation, or artifact assembly around provider work.
The configured minimum platform margin before quote floors, ceilings, or specific provider/workflow overrides.
How pricing works
Providers publish base rates; the platform applies a service fee for the control plane that turns those rates into reliable batch execution.
Provider base rate
The model catalog starts with the provider's token rate before service fees. These are the rates shown on Models.
Service fee
The platform adds a service fee to cover routing, orchestration, retries, fallback handling, receipts, and provider settlement operations.
Accepted quote
The customer accepts the final quote before work is created. The accepted quote is locked for that batch unless it expires before submission.
What the service fee covers
The fee pays for work the platform performs around the provider call, not for the raw model tokens alone.
Routing and quote locks
The platform evaluates providers, capacity, privacy fit, SLA fit, health, sanctions, price policy, and model compatibility before a quote is accepted.
Batch execution control
The platform coordinates dispatch, polling, retries, staged OpenWeight fallback, OpenRouter rescue, artifact assembly, and webhook delivery.
Marketplace operations
The fee funds provider onboarding, credential checks, provider earnings accounting, payout statements, and audit trails.
Where customers see it
The public catalog stays clean: provider rates first, service economics in pricing and receipts.
Billing lifecycle
A batch only moves from estimate to settlement after the customer accepts the quote.
AI agents: No seat pricing. Agents pay only for what they use — per token, per request. Learn more about agent access.
