OpenAI: GPT-5

GPT-5 is a paid multimodal model from OpenAI that accepts text, images, and files as input. It supports a 400,000-token context window with up to 128,000 tokens of output per completion, and it has tool use and reasoning capabilities enabled. Structured output support is not confirmed in available data. At $1.25 per million input tokens and $10.00 per million output tokens, GPT-5 sits at a moderate input cost but a relatively high output cost, which matters for tasks that generate long responses. Its blended benchmark score of 82.0 is based on only two benchmarks, so treat that figure as indicative rather than comprehensive. Teams running code-heavy workloads may find the 88.0 score on the coding benchmark useful signal, but buyers who need broader, independently verified performance data should note that coverage here is limited.