zuplo 7.12.2 → 7.12.4

package/docs/ai-gateway/cookbooks/dynamic-model-routing.mdx

@@ -17,8 +17,9 @@ Two public primitives from `@zuplo/runtime` do the work:
 
 - `AIGatewayModels.load(context)` returns the cached provider catalog, including
   each model's capability, status, and token rate card (`tokenRateCard`): the
-  prices the gateway charges, which are the list prices with your project's
-  prices applied. Credentials are never returned.
+  prices the gateway charges, which are the list prices with that provider's
+  [custom prices](../custom-pricing.mdx) applied. Credentials are never
+  returned.
 - `AIGatewayModelRouting.set(context, routing)` validates the routing, resolves
   provider credentials internally, and stores the selection the AI Gateway
   handler uses. `AIGatewayModelRouting.get(context)` reads the current

package/docs/ai-gateway/custom-pricing.mdx

@@ -0,0 +1,96 @@
+---
+title: AI Gateway Custom Pricing
+sidebar_label: Custom Pricing
+description:
+  Set your own per-token prices on a built-in AI Gateway provider so cost
+  budgets, budget alerts, and spend analytics reflect the rates you actually
+  pay.
+---
+
+The AI Gateway prices every request from Zuplo's model catalog, which lists each
+vendor's published per-token prices. If you've negotiated different rates with a
+vendor—a committed-use discount, an enterprise contract, or credits—set those
+prices on the provider so the gateway uses what you actually pay.
+
+A price you set applies everywhere the gateway computes cost:
+
+- Cost [budgets](./usage-limits.mdx) and [budget alerts](./budget-alerts.mdx)
+- Spend in the gateway, pool, and app analytics
+- The `tokenRateCard` that
+  [custom policies](./cookbooks/dynamic-model-routing.mdx) read from
+  `AIGatewayModels.load`
+
+Prices apply to new requests within about a minute of saving. Setting a price
+doesn't redeploy your gateway.
+
+## What you can price
+
+You can set a price for any rate the catalog lists for a model on a built-in
+provider:
+
+| Rate                       | What it prices                                                |
+| -------------------------- | ------------------------------------------------------------- |
+| **Input**                  | Prompt tokens                                                 |
+| **Output**                 | Generated tokens                                              |
+| **Cache read**             | Input tokens served from the provider's prompt cache          |
+| **Cache write (5 minute)** | Input tokens written to a cache entry that lasts five minutes |
+| **Cache write (1 hour)**   | Input tokens written to a cache entry that lasts one hour     |
+
+Models that charge more above a prompt-size threshold also have long-context
+rates. Those keep Zuplo's list prices.
+
+A rate the catalog doesn't list for a model, such as a 1-hour cache write on a
+model without one, can't be priced.
+
+## How prices apply
+
+Prices belong to a provider, not to your whole project. Two providers for the
+same vendor can carry different prices: for example, `anthropic` and
+`anthropic-2` with API keys under two different contracts. The gateway prices
+each request at the rates of the provider named in its model reference, so
+`anthropic-2/claude-sonnet-5` uses the prices on `anthropic-2`.
+
+For each rate, the gateway uses your price when you've set one and Zuplo's list
+price otherwise. A rate you leave unset follows the list price, including future
+changes to it. Set only the rates your contract changes.
+
+Deleting a provider deletes its prices. A provider you create later with the
+same name starts at list prices.
+
+[Custom providers](./custom-providers.mdx) don't use the catalog. You enter
+their per-token costs on each model when you add it, so there's no list price to
+override.
+
+:::note{title="OpenRouter"}
+
+[OpenRouter](./openrouter.mdx) reports the cost of each request itself, and the
+gateway records that figure instead of computing one from prices. A price you
+set on an OpenRouter provider applies only to a request whose response carries
+no cost.
+
+:::
+
+## Set prices in the portal
+
+Setting prices requires the same **Edit** permission as
+[managing providers](./managing-providers.mdx).
+
+<Stepper>
+
+1. Open
+   [**Settings → AI Providers**](https://portal.zuplo.com/+/account/project/ai/settings/data-models)
+   in your AI Gateway project in the Zuplo Portal.
+
+1. Click **Edit** next to a built-in provider.
+
+1. In **Available Models**, find the model and enter your price in a price cell.
+   Prices are per 1M tokens. An empty cell shows Zuplo's list price as its
+   placeholder; leave it empty to use the list price.
+
+   A model with at least one price of your own shows a **Custom price** badge.
+   To return every rate on that model to list prices, click **Reset**. The
+   change takes effect when you save.
+
+1. Click **Save**.
+
+</Stepper>

package/docs/ai-gateway/custom-providers.mdx

@@ -96,7 +96,9 @@ To add a custom AI provider to your Zuplo AI Gateway, follow these steps:
 
 1. Finally, add the available models that are hosted with your custom provider,
    along with their type. Optionally, you can add a dollar cost for input and
-   output tokens if you wish to track this via the AI Gateway.
+   output tokens if you wish to track this via the AI Gateway. A custom
+   provider's models have no other prices. To change the prices of a built-in
+   provider's models, see [Custom Pricing](./custom-pricing.mdx).
 
 1. Click **Create**.
 

package/docs/ai-gateway/managing-providers.mdx

@@ -98,6 +98,10 @@ every `providerName/model` reference, so renaming it would orphan each stored
 reference to this provider. After making your changes, click **Save** to apply
 them.
 
+On a built-in provider, you can also set your own per-token prices for its
+models, so cost budgets and analytics use the rates you negotiated with the
+vendor. See [Custom Pricing](./custom-pricing.mdx).
+
 :::caution
 
 Be careful removing models from a provider that's in use as it may break

package/docs/ai-gateway/usage-limits.mdx

@@ -108,6 +108,9 @@ action:
 | **Limit**  | An absolute amount in the meter's own unit                        |
 | **Action** | **Block** rejects the request; **Warn** notifies without blocking |
 
+The **Cost** meter prices each request at the provider's rates: Zuplo's list
+prices, or your own where you've set [custom prices](./custom-pricing.mdx).
+
 Add as many rows as you need to cap several meters and periods at once. A Warn
 row and a Block row on the same meter and period give you warn-then-block—warn
 at $80, block at $100. The warn amount has to be the lower of the two.

package/docs/ai-gateway/vertex-ai.mdx

@@ -299,7 +299,8 @@ change is live once the deployment completes.
 The per-token prices shown in the model picker are recorded rates, not a live
 read of Google's price list, and some Gemini models are on promotional rates
 that Google marks as temporary. Treat the gateway's cost figures as an estimate
-and your Google Cloud bill as the record.
+and your Google Cloud bill as the record. If you pay different rates, set them
+as [custom prices](./custom-pricing.mdx).
 
 :::
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "zuplo",
-  "version": "7.12.2",
+  "version": "7.12.4",
   "type": "module",
   "description": "The official Zuplo CLI for local development and platform management",
   "homepage": "https://zuplo.com/docs/cli/overview",
@@ -32,9 +32,9 @@
     "zuplo": "zuplo.js"
   },
   "dependencies": {
-    "@zuplo/cli": "7.12.2",
-    "@zuplo/core": "7.12.2",
-    "@zuplo/runtime": "7.12.2",
-    "@zuplo/test": "7.12.2"
+    "@zuplo/cli": "7.12.4",
+    "@zuplo/core": "7.12.4",
+    "@zuplo/runtime": "7.12.4",
+    "@zuplo/test": "7.12.4"
   }
 }