Together AI
Hosted inference for open-weight models such as Llama, Mixtral and Qwen, plus fine-tuning, at a lower price than most proprietary-model APIs.
Tested on Llama 3.3 70B ·
SERVICE · Usage-based, pay per token; dedicated endpoints billed hourly
Together is an easy pick when the model you want is open-weight and you would rather not run it yourself. You get one API, a wide model catalogue, and prices that undercut hosted proprietary models by a real margin. Fine-tuning and dedicated endpoints are there for when a shared endpoint stops being enough.
It hosts open models, so the newest proprietary release is not here. That is the price of the lower price. If you need one specific model and cannot settle for an open model that is good enough, check the model list before you commit.
Notes from the author
Start on the shared serverless endpoint and move to a dedicated one only after you have real usage numbers. Dedicated pricing is a flat hourly rate whether you use it or not.
For information only, not professional advice. You use it at your own risk and we are not liable for any loss or damage. See the disclaimer.
Tagged