# LLM API Billing Rules Explained - Documentation

> For the complete documentation index, see [llms.txt](/llms.txt). Markdown is available with `Accept: text/markdown` and `.md` URL variants.

Source: /docs/changelog/27-03-26

# LLM API Billing Rules Explained

Dear novita AI users,
This notice clarifies our LLM API billing policy, with a focus on how disconnected requests (HTTP 499) are handled.

##

[​](#pricing)

Pricing

novita AI LLM API is billed per token:

- Input Tokens — tokens consumed by the prompt

- Output Tokens — tokens generated by the model

Total Cost = Input Tokens × Input Rate + Output Tokens × Output Rate
Model-specific rates are available on the [Pricing Page](https://novita.ai/pricing).

##

[​](#billing-conditions)

Billing Conditions

Charges apply when the model has begun inference. Requests that do not reach the model are not charged.

ScenarioChargedReasonSuccessful request (200)YesModel completed inferenceClient disconnected (499)YesModel had begun inferenceInvalid request / auth failure / rate limit (400/401/403/429)NoRequest rejected before reaching the modelError due to platform issues (500/503/504)NoInfrastructure error; absorbed by the platform

Full status code reference: [Error Codes](/docs/guides/error).

##

[​](#499-client-disconnected)

499 (Client Disconnected)

When a request reaches the model, inference starts immediately. If the client disconnects mid-request, the compute resources already consumed are still billable.

Request ModeBillingNon-StreamingCharged for actual token usage regardless of disconnect timingStreamingCharged for actual token usage regardless of disconnect timing

This policy is consistent with standard industry practice across major LLM providers.

##

[​](#recommendations)

Recommendations

- Set `max_tokens` to control output length at the request level

- Configure client timeout ≥ 60s to prevent unintended disconnections

- Select appropriate models based on task requirements to optimize cost

##

[​](#faq)

FAQ

Why am I charged if I disconnected before receiving a response?

Inference begins upon request arrival. Disconnecting does not stop computation already in progress.

How can I avoid 499 charges?

Use the `max_tokens` parameter to cap output length before sending the request.

Where do 499 charges appear in my account?

In the usage dashboard, listed alongside standard requests with actual token consumption.

For further questions, please contact our support team.
The novita AI Team

Last modified on March 31, 2026