
Unlocking Creativity with Azure OpenAI
By :

The Provisioned Throughput Unit (PTU) feature lets you set the throughput you want for your application. It gives you more control over how you use and configure OpenAI’s large language models at a scale. It provides a dedicated compute to OpenAI models with a guaranteed throughput. You can set the total number of throughput units (PTU) you want and have the ability and control to distribute your commitment to OpenAI model you prefer. Each model needs a different amount of PTUs to run, for example GPT-3.5 needs less amount of PTUs compared to GPT4. You can select from various commitment options. With a 1-month or 1-year commitment, you can secure provisioned throughput and get savings in pricing. The provisioned throughput model offers more control and flexibility over workload needs, ensuring that the system is ready when higher workloads arise.
This feature enables: