I'm looking for some help in putting together a budget estimate for deploying a private version of DeepSeek, specifically R1 and V3. This setup needs to support a company with about 3,500 employees. Any insights on what the costs might look like would be greatly appreciated!
4 Answers
I found this article which might help you in understanding the local deployment costs: https://www.aifun.cc/en/deepseek-local-deployment-cost.html.
Based on my experience at OpenRouter, DeepInfra and Novita are solid options. For a company of your size, I’d recommend connecting directly to the DeepSeek API to manage costs better.
You might want to reach out to DeepInfra or other AI providers. Several of them offer private deployment solutions that could fit your needs.
Thanks for the suggestion!
If you're considering hardware too, investing in a couple of Mac Studios with 512GB of VRAM could be a good start for your deployment.
OP meant private inference, though, so API might not be the best route.