The ChatGPT API
The ChatGPT API, powered by OpenAI's "gpt-3.5-turbo" AI model, is priced at $0.002 per 1,000 tokens, or about 750 words. Early adopters include Snap, Quizlet, Instacart, and Shopify.
Greg Brockman, the president and chairman of OpenAI, says that the API will make the technology "much more usable and accessible." He added that the company is moving towards a "higher-level API" with Chat Markup Language (ChatML), allowing developers to better tailor and filter the ChatGPT model's responses.
Dedicated Capacity Plans
OpenAI has also launched dedicated capacity plans, giving large enterprises deeper control over system performance. Customers will be able to pay for an allocation of compute infrastructure to run an OpenAI model and will have full control over the instance's load. They will also be able to enable features such as longer context limits, which could lead to less hallucination by the model.
Brockman says that dedicated capacity customers can expect gpt-3.5-turbo models with up to a 16k context window, allowing a much larger amount of text to be considered. He also hinted at a general release in the future, but only sometime soon.
Potential Issues
Early on, ChatGPT was prone to racist and sexist answers due to its training data, which included internet content such as ebooks, Reddit posts, and Wikipedia articles. The technology is also susceptible to malicious adversarial prompts. However, OpenAI has taken steps to prevent these unintended behaviors, including more frequent model updates and the use of ChatML.
