GPT (OpenAI model family)

GPT (Generative Pre-trained Transformer) is OpenAI’s flagship family of general-purpose language and multimodal models. The line traces back through GPT-2 (2019) and GPT-3 (2020) to GPT-4 (2023), and has since advanced to a GPT-5 generation. The models power both the OpenAI API and consumer products such as ChatGPT.

As of June 2026, OpenAI’s developer models page lists a GPT-5 series as the current general-purpose lineup, naming GPT-5.5 and GPT-5.4 along with smaller GPT-5.4-mini and GPT-5.4-nano variants. The page also lists specialized GPT-branded models for images (gpt-image-2), realtime audio (gpt-realtime-2 and related variants), and an open-weight line marketed as gpt-oss. Specific version numbers move quickly; the names here reflect what the page showed on the verification date.

Distribution is primarily commercial API access plus the ChatGPT applications, with the gpt-oss line offered as downloadable open weights per the same page.

Why business readers should care: GPT models are the most widely deployed commercial AI models, and the named flagship on OpenAI’s own page is the most reliable signal of what your vendors and competitors can actually buy and build on today.

Sources

Last verified June 6, 2026