GPT-3 Models by OpenAI

OpenAIProprietary

OpenAI releases · 46 in the last 12 monthsChangelog →

This model family is considered obsolete. Consider newer alternatives in Related Model Families below.

4 models2020–2023Up to 2K ctxFrom $0.400/1M input

Last refreshed 2026-10-17. Next refresh: weekly.

Details

ResearcherOpenAI

LicenseProprietary

Commercial useRestricted

Models4

Released2020–2023

Max context2K

Links

Website

About

GPT-3, or Generative Pre-trained Transformer 3, is a large language model developed by OpenAI with a staggering 175 billion parameters, making it one of the largest AI models to date 12. This scale allows it to perform a wide range of tasks, from generating human-like text to translating languages and composing creative content 29. GPT-3 encompasses a family of models varying in size, all underpinned by a transformer architecture 1. Although initially accessible to developers broadly, Microsoft now holds exclusive licensing rights to its core technology, though OpenAI continues to provide API access 43. Despite its impressive capabilities, GPT-3 has faced criticism regarding biases arising from its training data and its inability to truly understand content, issues that subsequent models aim to resolve 210.

Decision facts

Best fit: General model comparison
Capability starting point: babbage with 2K context
Lowest tracked input: babbage · $0.400/1M · Azure OpenAI
Closest related family: GPT-5

Archived Variants

Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.

4 in view4 retired

babbageArchived

Keep only for existing workloads; choose a current variant for new builds.

2023-082K context1.3B parameters

davinciArchived

Keep only for existing workloads; choose a current variant for new builds.

2023-082K context175B parameters

adaArchived

Keep only for existing workloads; choose a current variant for new builds.

2020-062K context

curieArchived

Keep only for existing workloads; choose a current variant for new builds.

2020-062K context6.7B parameters

Current GPT-3 variants with use-when guidance and lifecycle status
Model	Use when	Released	Signals	Status
babbage	Keep only for existing workloads; choose a current variant for new builds.	2023-08	2K context1.3B parameters	Archived
davinci	Keep only for existing workloads; choose a current variant for new builds.	2023-08	2K context175B parameters	Archived
ada	Keep only for existing workloads; choose a current variant for new builds.	2020-06	2K context	Archived
curie	Keep only for existing workloads; choose a current variant for new builds.	2020-06	2K context6.7B parameters	Archived