OpenAI today released GPT-5 Turbo, a significantly upgraded model featuring native multimodal reasoning that can seamlessly process text, images, audio, and video within a single inference pass without separate encoding steps.
Benchmark results show GPT-5 Turbo achieves 92 percent accuracy on the MMLU-Pro evaluation, a 15-point improvement over its predecessor. The model also demonstrates substantially improved performance on mathematical reasoning, code generation, and long-context document analysis tasks.
API pricing is set at $5 per million input tokens and $15 per million output tokens, a 40 percent reduction from GPT-5 standard pricing. OpenAI is offering the model through a new batch processing API that further reduces costs for non-real-time workloads.
