OpenAI

About OpenAI

The OpenAI API provides programmatic access to AI models developed by OpenAI, offered through a general-purpose "text in, text out" interface that works on a range of English-language tasks. Given a text prompt, the API returns a text completion matching the pattern of the input; it can be directed by providing a few examples within the prompt, and performance on specific tasks can be refined by training on a dataset of examples or through learning from human feedback. At launch, the API ran models from the GPT-3 family.

The API is intended for integration into products, development of new applications, and general experimentation with the technology, targeting both individual developers and machine learning teams. Access was limited to a private beta at launch, with usage subject to a mandatory production review process; OpenAI stated it would terminate access for use cases involving harassment, intentional deception, radicalization, astroturfing, or spam, or applications lacking sufficient safeguards against misuse.

Early users in the private beta included Algolia, Quizlet, and Reddit, along with researchers at institutions such as the Middlebury Institute, University of Washington, and the Allen Institute for AI through an academic access program.

Key features

• General-purpose text-in, text-out API interface for language tasks
• Few-shot prompting: customize behavior by showing example inputs/outputs
• Fine-tuning on custom datasets to improve task-specific performance
• Runs models from the GPT-3 family with ongoing speed and throughput improvements
• Mandatory production review process to vet applications before going live

Frequently asked questions