GPT-4 Turbo is a large multimodal model developed by OpenAI, representing an optimization of the GPT-4 architecture. It was designed to offer higher performance and efficiency compared to original GPT-4 variants, featuring a significantly expanded context window and more recent world knowledge. The model was initially introduced as a preview in November 2023 before reaching general availability.
The model supports a 128,000-token context window, enabling it to process the equivalent of over 300 pages of text in a single prompt. This capacity allows for the analysis of extensive documentation, long-form codebases, and complex instructions without the need for extensive retrieval-augmented generation (RAG) or chunking. Its training data knowledge cutoff was updated to April 2023 for initial versions and December 2023 for the final production-ready release.
Capabilities and Multimodality
GPT-4 Turbo is capable of processing both text and visual inputs. Through its vision-enabled versions, the model can perform tasks such as describing image contents, interpreting technical diagrams, and extracting text from documents. It also introduced specialized features for developers, including a JSON mode for ensuring structured outputs and improved instruction following for complex task execution.