Rejecting this for a few reasons:

Performance benchmarking against third-party services strongly discouraged. Results can be biased by network variability, rate limits, and other external factors. Additionally it is often against ToS to benchmark a service. For example, looking through the d.run ToS I found:

4.1.6 [...] using automated tools, scripts, or other means to send large quantities of requests or consume network bandwidth that severely impacts the d.run or other users' normal use. We reserve the right to limit your service access due to violations of these fairness principles, including but not limited to suspending or terminating services.

Finally, we do not want to promote any AI service vendor in our official documentation. We're open to bug fixes and enablement for hosted AI services, but GuideLLM is intended for use with your own infrastructure; other use-cases will not be officially endorsed or supported.

Rejecting this for a few reasons:

Performance benchmarking against third-party services strongly discouraged. Results can be biased by network variability, rate limits, and other external factors. Additionally it is often against ToS to benchmark a service. For example, looking through the d.run ToS I found:

4.1.6 [...] using automated tools, scripts, or other means to send large quantities of requests or consume network bandwidth that severely impacts the d.run or other users' normal use. We reserve the right to limit your service access due to violations of these fairness principles, including but not limited to suspending or terminating services.

Finally, we do not want to promote any AI service vendor in our official documentation. We're open to bug fixes and enablement for hosted AI services, but GuideLLM is intended for use with your own infrastructure; other use-cases will not be officially endorsed or supported.

If i adopt vllm-simulator in one docker container [https://llm-d.ai/docs/architecture/Components/inference-sim] to start up one vllm backend server to describe benchmark process, is that ok?

Yes!! That sounds perfect.