Latest GPT Developments: GPT-4 Turbo and New Features Released

Recently, OpenAI has released a series of significant updates and new features for the GPT models.

Firstly, GPT-4 Turbo, as the latest model preview, has been launched with significant improvements. Compared to previous versions, GPT-4 Turbo has made significant advancements in code generation, task completion, and handling non-English UTF-8 characters. The new version has expanded the context window to 128K tokens and comes at a lower price, greatly reducing the cost of usage.

Additionally, GPT-4 Turbo now supports visual input, allowing it to accept image inputs in the Chat Completions API. This feature enables applications such as generating image descriptions, analyzing real-world images, and reading documents with charts. For example, the BeMyEyes project utilizes this technology to assist visually impaired individuals with daily tasks like product recognition or navigating stores.

OpenAI has also introduced new embedding models, text-embedding-3-large and text-embedding-3-small, which maintain conceptual representation capabilities while allowing embedding dimensions to be adjusted through API parameters, striking a balance between performance and cost. This flexibility enables developers to use these models in various storage and computing environments.