GPT-4 API Is Here: The Future of AI with OpenAI

OpenAI introduces a shift towards GPT-4 API, promising a new wave of AI possibilities. Explore what this means for developers and the AI industry

Friday July 07, 2023 , 3 min Read

In a dynamic demonstration of forward-thinking innovation, OpenAI, a pioneer in the AI realm, has unveiled the pivotal role that GPT-4 API will play in its strategic approach, as of July 7, 2023. This stride toward the GPT-4 API release coincides with a systematic plan to phase out older models in the Completions API. The move mirrors OpenAI's enduring commitment to providing developers with an ever-evolving, user-friendly platform, powered by cutting-edge technologies.

As 2024 dawns, OpenAI intends to sunset older models within the Completions API. This strategic step aims to streamline computational resources while redirecting the focus toward the Chat Completions API. Despite these changes, developers can still expect a seamless transition and uninterrupted access to the Completions API, albeit under a newly designated “legacy” category.

GPT-4 API is now available to all paying OpenAI API customers. GPT-3.5 Turbo, DALL·E, and Whisper APIs are also now generally available, and we’re announcing a deprecation plan for some of our older models, which will retire beginning of 2024: https://t.co/H11w4IalNZ
— OpenAI (@OpenAI) July 6, 2023

From January 4, 2024, existing completion models will give way to their more advanced, efficient successors. Developers utilising GPT-3 models, known under their respective code names of ada, babbage, curie, and davinci, can anticipate an automatic upgrade to the new iterations: ada-002, babbage-002, curie-002, and davinci-002.

OpenAI is keenly aware of the potential challenges these changes may bring, and has implemented detailed strategies to minimise disruptions. Developers currently using older completion models are advised to manually update their integrations, specifying the new models within their API requests. To ease this transition, OpenAI has introduced compatible replacements, such as gpt-3.5-turbo-instruct, designed to seamlessly replace older models.

Interestingly, OpenAI is also addressing the needs of developers who have customised their models. Those developers will be prompted to adjust their fine-tuned models using the new base GPT-3 models or explore newer offerings like gpt-3.5-turbo, and gpt-4. To alleviate any potential difficulties, OpenAI is extending priority access to GPT-3.5 Turbo and GPT-4 fine-tuning to those who had previously adjusted older models, demonstrating a profound understanding of the complexities associated with migrating from models that have been intricately tailored to specific data sets.

The transition also affects users of older embedding models who will need to switch to text-embedding-ada-002 by the same January deadline. This new model, launched in December 2022, has demonstrated superior performance and cost-efficiency compared to its predecessors, accounting for an impressive 99.9% of all embedding API usage.

OpenAI's announcement also signals the end for the Edits API and its related models. Originally designed as an experimental tool for developers to revise prompts based on instructions, the Edits API will be supplanted by gpt-3.5-turbo and the Chat Completions API.

The shifts reflect OpenAI's firm dedication to its users, prioritising the delivery of sophisticated, economical, and accessible developer tools. These impending changes echo OpenAI's commitment to fostering innovation, equipping developers with state-of-the-art resources to navigate the fast-paced world of AI. OpenAI's bold transition heralds their relentless pursuit of breaking new ground in AI technologies.