Approved launch pricing
The approved plan catalog balances useful customer capacity with disciplined infrastructure, retention, and support boundaries. Feature availability follows Token Pilot's early-access product status and staged release process.
For founders and teams operating one AI application
For teams managing several AI workloads
For agencies managing AI economics across clients
For larger deployments, governance, and infrastructure needs
Your model relationships, API accounts, and provider billing remain yours.
Token Pilot pricing covers visibility, optimization intelligence, workflow controls, reporting, and support.
Usage alerts, manually approved request packs, and plan upgrades are preferred over interrupting normal provider traffic.
Plan comparison
These limits define the launch catalog used by the public site and the server-owned billing policy.
| Entitlement | Starter | Professional | Agency | Enterprise |
|---|---|---|---|---|
| Managed AI applications | 1 | 5 | 25 | Custom |
| Internal team users | 3 | 10 | 25 | Custom |
| Client viewer seats | — | — | 50 | Custom |
| Monthly AI requests observed | 250,000 | 2 million | 10 million | Committed volume |
| Detailed telemetry retention | 30 days | 90 days | 180 days | 365 days or custom |
| Provider connections | 3 | 15 | 75 | Custom |
| Client workspaces | — | — | 10 | Custom |
| Support target | 2 business days | 1 business day | 8 business hours | Contractual |
| Included onboarding | Self-service | One session | Two sessions | Custom |
| Account review | — | — | Quarterly | Custom |
Agency client-workspace and viewer-seat capabilities remain subject to staged early-access availability. Enterprise commitments require a written agreement.
Early-access usage packs
Request packs increase Token Pilot's approved observation allowance. They do not include or resell model-provider tokens and are manually approved during early access.
Pricing questions
No. You keep your provider accounts and pay model providers directly. Token Pilot does not add a token-credit markup.
One AI request observed or processed through Token Pilot counts toward the plan threshold. Dashboard views, internal API calls, and ordinary database operations do not count as model requests.
One provider credential connected to one managed application counts as one connection. Connecting one application to OpenAI, Anthropic, and Gemini uses three connections.
Token Pilot is designed to alert before the threshold and preserve normal provider traffic. During early access, additional usage is handled through a manually approved request pack or plan upgrade rather than an automatic hard shutdown.
The stated limit applies to detailed request telemetry. Billing, entitlement, security, and immutable audit records may follow separate retention requirements.
Yes. Annual agreements can include preferred pricing and onboarding terms based on the plan and deployment.
No. Token Pilot identifies and measures opportunities, but actual results depend on workload behavior, provider pricing, technical eligibility, and approved changes.