Skip to content

Enterprise Pricing ​

By default, every TokenSG account is billed at list price. If you have signed an enterprise agreement or joined a partner/channel program, we configure negotiated pricing on your account — same models, same API key, billed at your contracted rate.

This page explains where negotiated pricing applies, where you can see it, and which price your invoice is based on.

At a glance: where negotiated pricing shows up ​

SurfacePrice shown
Dashboard (signed in) — model list / price comparison / model detailYour negotiated price
Dashboard (signed out), and the model/pricing pages on tokensg.aiList price
The pricing field in GET /v1/modelsList price (see below)
Dashboard usage logs and invoicesYour negotiated price (the price you are actually billed at)

/v1/models returns list price, not your negotiated price

GET /v1/models is the OpenAI-compatible endpoint. It returns the public price catalog and does not vary by caller — using it for programmatic cost estimation will overstate your cost (or, in rare uplift cases, understate it).

Billing always uses your account's negotiated price. Treat the dashboard usage logs and invoices as the source of truth: every request log records the discount rate that applied, which rule it matched, and the equivalent cost at list price for comparison.

How negotiated pricing applies ​

  • Per account. Negotiated pricing is attached to your TokenSG account, and every API key under that account shares it. No key rotation, no base_url change, no extra headers.
  • Three scopes, matched from most to least specific — the first match wins:
    1. Model — a single model (e.g. claude-sonnet-4-20250514)
    2. Provider — every model from one provider
    3. Account-wide — every model the account calls
  • Multiple accounts (e.g. several teams in one company) can be placed in a shared discount group so they all get the same rates. If an individual account also has its own rule, the account's own rule wins at that scope — rules override, they do not stack or multiply.
  • Time windows are supported for promotional or quarterly rates. A window is half-open (start inclusive, end exclusive); once it ends, pricing falls back to list price automatically.

Model IDs match exactly — aliases do not share a discount

Discounts are matched on the exact model ID. A date-pinned ID and its rolling alias (for example claude-haiku-4-5 and claude-haiku-4-5-20251001) are two separate models in the catalog and are priced independently.

If your application sends both forms, confirm with us that your agreement covers the exact model ID you send — anything not covered is billed at list price.

How savings are calculated ​

The "total saved" figure on the dashboard is official price − amount paid, where the official price is the model's public rate at the original provider. When negotiated pricing applies, the amount paid already reflects your contracted rate, so your savings automatically include it. Nothing to configure.

Each individual request log shows:

  • Applied rate — the ratio actually used for that request
  • Matched rule — model / provider / account-wide, and whether it came from a discount group
  • Cost at list price — what the same request would have cost at list price; the difference is what negotiated pricing saved you

FAQ ​

A colleague and I see different prices — why? Negotiated pricing is per account. Other accounts (and signed-out visitors) see list price. This is expected.

Prices changed after I signed out. Signed-out pages show the public price catalog. Sign back in and your negotiated prices return.

Can a negotiated price be higher than list price? Yes, in arrangements involving dedicated capacity, reserved channels, or a separate SLA. In that case the dashboard does not present it as a discount and does not show the "negotiated price" marker — pricing and invoicing follow the agreement you signed.

How do I request negotiated pricing? Talk to your customer success contact, or send an enquiry from the Dashboard. Once it is configured, no code change is needed — your very next request is billed at the new rate.