Early Access Notice and Limitations

With the release of version 0.3 of our OuteTTS model, we are excited to announce the launch of early access to our services. This initiative is designed to support the development and creation of open-source models. During this phase, we welcome your feedback to refine and enhance our services.

Early Access Service Limitations

To ensure a smooth experience for all users, some temporary limitations are in place during the early access phase:

• Request Limit: Each user is limited to 100 requests per day across both the API and Playground.
• Playground Storage: The Playground can store up to 10 assistant-side messages (generated audio responses). Older messages will be automatically removed as new ones are generated.
• Dynamic Resource Sharing: Infrastructure uses globally shared resources with an inactivity timeout. This may lead to "cold starts" on GPUs, which can result in initial request delays of approximately 1 minute when a GPU is not already active.
• Stability: As this is an early access release, occasional bugs may occur in the Playground and API.

We appreciate your understanding as we fine-tune these systems to provide a reliable and efficient service.

Model Considerations

The OuteTTS model is still under active development. Users may occasionally encounter unexpected outputs or inaccuracies in responses.

Pricing

OuteTTS 0.3 500M is currently available via our API and Playground at a rate of $0.01 per generated audio second.

Costs are calculated based on the final audio file or the generated tokens, depending on the specifics of the request and how it was handled.

• For partially completed requests, usage charges will apply up to the point of cancellation.
• In the Playground, the system will attempt to store any partial audio generated for your convenience.
• When using the API, partial results cannot be stored, but associated charges will still reflect the resources used.

These measures ensure fair usage and help maintain the availability of our services for all users. We encourage you to consider these details as you plan your requests.