Hi Garry,

I hope this message finds you well.

We are a startup currently engaged in R&D focused on decentralized inference. As part of our closed beta, we're exploring various options for integrating paid solutions with our self-hosted cluster. TogetherAI stands out as a great option, particularly due to your competitive pricing per million tokens.

Given that we are still uncertain about demand, we believe a serverless endpoint for the Llama-405B base model would be more cost-effective than dedicated instances at this stage. Could you assist us in setting up a dedicated endpoint on a serverless architecture for this model?

We look forward to your guidance and assistance.

Best regards,

Michael

Got it - we will make it available for dedicated endpoints in the next week or so, but won't make it available for serverless.