GPT-4o (August 2024), identified by the version snapshot gpt-4o-2024-08-06, is a multimodal generative model developed by OpenAI. This specific iteration was released to enhance the model's reliability in programmatic environments, offering improved reasoning and instruction-following capabilities over previous snapshots.
A major addition in this release is Structured Outputs, a feature designed to ensure that model-generated responses strictly conform to developer-defined JSON schemas. This capability addresses formatting issues found in earlier models, enabling more consistent integration into software workflows and achieving high reliability in schema-following benchmarks.
The model supports a context window of 128,000 tokens and can generate up to 16,384 output tokens in a single request. Along with performance improvements, the August 2024 update introduced cost optimizations for API usage and expanded support for model fine-tuning, allowing developers to customize the model using proprietary datasets.