1. The service
The vllm-sr/auto API at mom.vllm-sr.ai and pytorch.vllm-sr.ai (“we”) is a free service, offered for PyTorch Conference and the open-source community, that answers OpenAI-compatible chat requests with open models chosen per request by vLLM Semantic Router. Models, limits, features, speed and availability may change at any time, and there is no performance guarantee. By creating an API key or using the API, you agree to these terms.
2. Eligibility and keys
You must be at least 18 years old to create a key or use the service. By creating a key, you confirm that you are.
You are responsible for your keys and for everything sent with them. Keep them secret: do not post them publicly, put them in client-side code or public repositories, or use someone else’s key without permission. A key is shown once and cannot be recovered; you can revoke it at any time.
3. Acceptable use
Do not use the service to break the law, infringe the rights of others, compromise security, abuse the infrastructure or disrupt the service for other users. Do not try to get around rate limits, quotas, key-creation limits, authentication or any other control, for example by creating many keys. Do not send content you have no right to send.
4. Quotas and limits
Every key has rate limits, a concurrency limit and daily quotas, and every request has size limits; the current values are in the docs and at /pt/api/info. They may change at any time, based on demand, capacity, abuse prevention and operational needs. High-volume, automated or risky use may be limited, delayed or blocked without notice.
5. Logging and data use
Requests and responses, including images, are logged, stored and processed to operate, secure, debug, analyse and improve the service, including how requests are routed, and to train and evaluate models. No request header turns this off.
We sanitise text before analysis where feasible, to reduce sensitive or identifying content. Sanitisation does not guarantee that all sensitive information is removed. De-identified data derived from requests, such as sanitised requests and responses and usage statistics, may be published or open-sourced.
Do not send sensitive personal information, confidential information, regulated data, secrets or credentials, or data you are not authorised to process. When you send content, you confirm that you may share it for these purposes.
IP addresses are kept for up to 30 days, for abuse prevention. You can revoke a key with DELETE /pt/api/keys/self; its stored data is kept. To revoke a key and delete everything stored for it, use POST /pt/api/keys/self/forget, or Delete key and its data on the Keys page. Requests sent with the shared workshop key are logged the same way; that key is not personal, so its data cannot be deleted per person. The privacy notice has the details.
6. No warranty
The service is provided as is and as available, without warranty of any kind. Answers may be wrong, incomplete, delayed or unavailable; check them before you rely on them. Do not rely on the service for legal, medical, financial, safety-critical or other high-risk decisions. To the extent the law allows, we are not liable for any damage that results from using the service or from being unable to use it.
7. Suspension and termination
We may limit, suspend or end access, revoke or disable keys, and change, restrict or remove models and features at any time, for misuse, security, operational needs, a breach of these terms, or the end of the service.
8. Changes
We may update these terms. The current version is always on this page, and using the service after a change means you accept the new version.
9. Contact
Questions about the service or these terms go to the vLLM Semantic Router issues on GitHub. Never post a key or the content of a request in an issue.