OpenAI introduces GPT-4 Turbo with increased context and affordable price for developers
OpenAI has announced improvements to its GPT-4 and GPT-3.5 language models. The new versions include updated knowledge bases and a longer context window.
GPT-4 Turbo, available as an API preview, has been trained on data through April 2023. The early GPT-4 used data only until September 2021. The full release of Turbo is scheduled for the coming weeks.
GPT-4 Turbo also sees more context – 128 KB of text at a time. This will allow the model to better understand the questions and give thoughtful answers. The previous versions had 8K and 32K context.
In addition, Turbo is cheaper for developers – $0.01 per 1000 tokens at the entrance and $0.03 at the exit. This is three times lower than the previous price, according to OpenAI.
GPT-4 will continue to accept text queries, picture prompts, and integrate DALL-E 3’s image generation feature. In one request, you can ask for the result in the desired format: XML, JSON, etc.
The smaller GPT-3.5 Turbo model with a 16 KB context window will get similar improvements. Its cost will be $0.01 at the entrance and $0.002 at the exit.
In addition, OpenAI has promised to compensate customers for the costs of copyright infringement lawsuits, similar to Microsoft and Google. This applies to ChatGPT Enterprise and the developer platform.
