Dedicated vs Serverless models

together.ai serverless models are a great starting point for anybody building or experimenting with AI. They allow you to get started in minutes, on the model of your choice.

Once your experiments get a bit more advanced, or you're ready to launch your project to a wider audience there are a number of advantages to using a dedicated endpoint to deploy your model. These advantages include:

• Consistent, predictable performance, unaffected by other users' load in our serverless environment

• No rate limits, with a high maximum load capacity

• More cost-effective under high utilization (note that dedicated endpoints are charged based on how long the hardware is reserved for, so will have a cost even when idle)

• Access to a broader selection of models