> ## Documentation Index
> Fetch the complete documentation index at: https://docs.jojapi.com/llms.txt
> Use this file to discover all available pages before exploring further.

# July 2026 Changelog

## Billable Objects

Billing is no longer a single credit pool. An API now charges in **billable objects** — Credits remains the default, and providers can define more: AI tokens, compute time, bandwidth, or any resource worth metering.

**For providers:**

* A new **Objects** tab in Studio manages your API's objects. Name, slug and dimension are fixed at creation; the slug becomes part of your API's header contract (`X-Jojapi-<slug>-Used`).
* Each object can carry a **default usage formula** — the metering rule endpoints inherit unless they define their own. A plain number works as a constant; a template computes the amount from each response.
* The endpoint **Billing** section replaces the old fixed/flexible credits toggle: add any of your objects to an endpoint and give each a single cost — a number (fixed per request), a formula (metered), or nothing (inherits the object default). An optional description explains the metering to consumers in plain words.
* **Plans are priced per object**: fixed-price plans set an included quota per object, pay-as-you-go plans define a graduated tier ladder per object. The plan form warns you when your endpoints consume an object the plan doesn't cover — subscribers of such a plan would be blocked on those endpoints.

Documentation: [Billable Objects](/studio/billable-objects)

## Transparent Per-Object Pricing

Consumers now see exactly how usage is counted, everywhere:

* Endpoint cards show a **billing box** with one row per object: the fixed amount, or the exact metering formula (hover the info icon) with a last-month average per request, plus the provider's description.
* API pricing pages list **included quotas per object** on fixed plans and a **tier table per object** on pay-as-you-go plans, with ranges spelled out ("First 1,000", "1,001 – 5,000", "Over 5,000") and a note explaining graduated pricing: each unit is billed at the rate of the tier it falls in.
* The pay-as-you-go **cost calculator** estimates per object and totals the plan.

## Per-Object Usage Tracking

* **Workspace → Subscriptions** shows one usage bar per included object on fixed plans ("1,200 / 5,000 AI Tokens") and per-object usage with spending on pay-as-you-go plans. Overshoot is highlighted in red with the real numbers.
* Providers get the same per-object view of their subscribers' usage in **Studio → Subscriptions**, and pay-as-you-go spending totals now sum every object's ladder.
* The gateway emits per-object [response headers](/consumers/response-headers): `X-Jojapi-<slug>-Used`, `-Remaining` and `-Total-Used`. The classic Credits headers are unchanged.

## Failed Requests Are Always Free

Responses with a 5xx status — and gateway-generated errors — now bill **0 on every object**, with no exceptions. Additionally, the gateway only ever bills amounts it can actually compute: a metering formula that fails or produces a non-numeric value charges nothing instead of falling back to an implicit 1.

Documentation: [Adjusting Usage from Your Server](/studio/adjusting-credit-usage)