OpenAI has announced GPT-5 Turbo, a major upgrade to its flagship language model that introduces native multimodal reasoning across text, images, audio, and video inputs simultaneously. The model processes all modalities through a unified architecture rather than separate encoders.
Benchmark results show GPT-5 Turbo outperforms its predecessor by 40 percent on complex reasoning tasks and achieves near-human performance on graduate-level science and mathematics examinations. The model also demonstrates significantly improved factual accuracy and reduced hallucination rates.
API access is available immediately with pricing set at $5 per million input tokens, a 60 percent reduction from GPT-5's launch pricing, reflecting OpenAI's strategy to drive widespread adoption.
