Are inference providers able to make any margins?

<!-- SC_OFF --><div class="md"><p>I run all sorts of models locally, but I only have a 3090ti. With vllm I’m able to batch and get 2-6 parallel streams in low context. That

<!-- SC_OFF --><div class="md"><p>I’m thinking of expanding an on-device inference SDk into a full blown AI inference platform and seeing more and more inference platform popping

<!-- SC_OFF --><div class="md"><p>Hey everyone,</p> <p>I have been looking into how people source compute for their Inference workloads (and in general). I wanted to understand

<!-- SC_OFF --><div class="md"><p><a href="https://x.com/gajesh/status/2044067335506669782">https://x.com/gajesh/status/2044067335506669782</a></p> <p>Disclaimer: Not affiliated

<table> <tr><td> <a href="https://www.reddit.com/r/LocalLLaMA/comments/1tlobqd/inference_provider_tiers_by_cachehit_rates_using/"> <img alt="Inference provider tiers by Cache-hit