OpenAI has announced the general availability of GPT-4, its latest language model. GPT-4 has been powering various applications since its release in March, including those focused on visual accessibility, wealth management, language preservation, and language learning. The model is now accessible to paying API customers.
Existing API developers with a payment history can access GPT-4 with an 8K context. This means that the model can generate output based on approximately 13 pages of text. OpenAI plans to increase rate limits and open access to GPT-4 for new developers by the end of July, depending on compute availability. The company also intends to make GPT-3.5 Turbo, DALL-E, and Whisper APIs generally available to developers.
In terms of depreciation, OpenAI has announced that GPT-3 and its variations using the Completions API will retire by January 4, 2024. While the API will still be accessible, it will receive a legacy label and be replaced by a new model. Applications using stable names for base GPT-3 models will automatically be replaced by their successors.
Additionally, developers using older embedding models will need to migrate to the new embedding model released by OpenAI in December 2022 by January 4, 2024. OpenAI claims that the new model is more capable and cost-effective than its predecessors. The company acknowledges that the switch may be costly and inconvenient for some developers and has committed to covering any associated costs.
OpenAI’s GPT-4 represents a significant advancement in language models and offers exciting possibilities for various industries.
