Completion

Capable of very simple tasks, usually the fasted model in the gpt-3 series, and lowest cost. Can be fine-tuned. Has a context window of 2,049 tokens and training data up to October 2019.

Capable of straightforward tasks, very fast, and lower cost. Can be fine-tuned. Has a context window of 2,049 tokens and training data up to October 2019.

Optimized for code-completion tasks. Will be deprecated on Jan 4th

Very capable, but faster and lower cost than davinci. Can be fine-tuned. Has a context window of 2,049 tokens and training data up to October 2019.

Most capable gpt-3 model, can do any task the other models can do, often with higher quality. Can be fine-tuned. Has a context window of 2,049 tokens and training data up to October 2019.

Similar to text-davinci-003 but compatible with the legacy completions endpoint.

Capable of very simple tasks, usually the fastest model in the gpt-3 series, and lowest cost. Has a context window of 2,049 tokens and training data up to October 2019.

Capable of straightforward tasks, very fast, and lower cost. Has a context window of 2,049 tokens and training data up to October 2019.

Vary capable, faster and lower cost than davinci. Has a context window of 2,049 tokens and training data up to October 2019.

Similar capabilities to text-davinci-003 but trained with supervised fine-tuning instead of reinforcement learning. Will be deprecated on January 4th 2024. Has a context window of 4,096 tokens and training data up to June 2021.

Can do language tasks with better quality and consistency than the curie, babbage, or ada models. Will be deprecated on January 4th