GPT-4

GPT-4 is a multimodal large language model (LLM) developed by OpenAI and released on 14 March 2023 as the successor to GPT-3.5. It accepts text and image inputs and produces text outputs.

GPT-4 was offered with an 8,192-token context window and an extended 32,768-token variant. OpenAI did not disclose the parameter count or architecture details; the model is based on the Transformer architecture and was tuned using reinforcement learning from human feedback (RLHF).

The model achieved strong results on professional and academic exams — including roughly the 90th percentile on a simulated Uniform Bar Exam — and high scores on the MMLU, HumanEval and GSM8K benchmarks. Model weights remain closed; access is provided through the OpenAI API and Azure OpenAI, and historically through ChatGPT (Plus).