• btc = $67 789.00 569.16 (0.85 %)

  • eth = $3 245.01 0.80 (0.02 %)

  • ton = $6.68 0.00 (0.03 %)

  • btc = $67 789.00 569.16 (0.85 %)

  • eth = $3 245.01 0.80 (0.02 %)

  • ton = $6.68 0.00 (0.03 %)

7 Jul, 2023
1 min time to read

OpenAI has officially made its GPT-4 text-generating model available through its API.

The announcement comes after months of anticipation and requests from millions of developers. With GPT-4, developers can generate text, including code, and utilize image and text inputs, making it more versatile than its predecessor, GPT-3.5.

OpenAI has trained GPT-4 using a combination of publicly available data and licensed data, achieving human-level performance on various professional and academic benchmarks.

However, OpenAI acknowledges that GPT-4 is not perfect. It may sometimes generate incorrect facts or reasoning errors, and it lacks the ability to learn from its experiences. For instance, it may introduce security vulnerabilities into generated code. Despite these limitations, OpenAI sees great potential for chat-based models like GPT-4 to support various use cases.

OpenAI also revealed plans to allow developers to fine-tune both GPT-4 and GPT-3.5 Turbo, another recent text-generating model, with their own data. This capability, currently available for other OpenAI models, is expected to be introduced later this year.

In addition to GPT-4, OpenAI has made its DALL-E 2 image-generating model and Whisper speech-to-text model available through APIs. However, the image-understanding capability of GPT-4 is currently limited to a partnership with Be My Eyes.

OpenAI also announced that certain older models, including GPT-3 and its derivatives, will be replaced with new "base GPT-3" models starting January 4, 2024. Developers using the old models will need to upgrade their integrations, and those who wish to continue using fine-tuned old models should do so.

"We will be providing support to users who previously fine-tuned models to make this transition as smooth as possible," OpenAI wrote. "In the coming weeks, we will reach out to developers who have recently used these older models, and will provide more information once the new completion models are ready for early testing."