GPT-4 Models by OpenAI

OpenAIProprietaryHighlight
OpenAI releases · 48 in the last 12 monthsChangelog →
5 models2023–2024Up to 128K ctxFrom $5.00/1M input

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

Details

ResearcherOpenAI
LicenseProprietary
Commercial useRestricted
Models5
Released2023–2024
Max context128K

Capabilities

VisionAll models
Multimodal3 of 5 models
JSON / Tool use2 of 5 models
Structured Outputs3 of 5 models
Code ExecutionAll models

Links

Website

About

GPT-4 is a large multimodal model that accepts text or image inputs and outputs text. It can solve complex problems with greater accuracy than any of our previous models, thanks to its extensive general knowledge and advanced reasoning capabilities.

Decision facts

Best fit
vision and multimodal workJSON / Tool usestructured outputs
Capability starting point
GPT-4 Turbo with 128K context and JSON / Tool use, structured outputs, and multimodal inputs
Lowest tracked input
GPT-4 Turbo · $5.00/1M · Replicate API
Closest related family
Ember

Archived Variants

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

5 in view5 retired
GPT-4 TurboReplaced

Keep for legacy integrations; evaluate GPT-6.1 Sol before new work.

2024-04128K contextJSON / Tool usestructured outputs

Keep for legacy integrations; evaluate GPT-6.1 Sol before new work.

2023-11128K contextstructured outputscode execution

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

2023-11128K contextcode executionmultimodal inputs
GPT-4Replaced

Keep for legacy integrations; evaluate GPT-6.1 Sol before new work.

2023-038K contextJSON / Tool usestructured outputs
GPT-4 32kArchived

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

2023-0332K contextcode executionmultimodal inputs

Release Timeline

3 release groups
2024-04
1 retired
GPT-4 Turbo
128K contextJSON / Tool usestructured outputs
Replaced
2023-11
2 retired
GPT-4 Turbo Preview
128K contextstructured outputscode execution
Replaced
GPT-4 Vision Preview
128K contextcode executionmultimodal inputs
Archived
2023-03
2 retired
GPT-4
8K contextJSON / Tool usestructured outputs
Replaced
GPT-4 32k
32K contextcode executionmultimodal inputs
Archived

Replaced By

Keep for legacy integrations; evaluate GPT-6.1 Sol before new work.

Keep for legacy integrations; evaluate GPT-6.1 Sol before new work.

Keep for legacy integrations; evaluate GPT-6.1 Sol before new work.

Specifications(5 models)

GPT-4 model specifications comparison
ModelReleasedContextParametersVisionMultimodalJSON / Tool useStructured OutputsCode Exec
GPT-4 Turbo2024-04128K1.76T (8x222B MoE)*YesYesYesYesYes
GPT-4 Turbo Preview2023-11128K1.76T (8x222B MoE)*YesNoNoYesYes
GPT-4 Vision Preview2023-11128K1.76T (8x222B MoE)*YesYesNoNoYes
GPT-42023-038K1.76T (8x222B MoE)*YesYesYesYesYes
GPT-4 32k2023-0332K1.76T (8x222B MoE)*YesNoNoNoYes

Pricing

GPT-4 model pricing by provider
ModelProviderInput / 1MOutput / 1MType
GPT-4 TurboReplicate API$5$15Serverless
GPT-4 TurboOpenAI API$10$30Serverless
GPT-4 Turbo PreviewOpenAI API$10$30Serverless
GPT-4 TurboAzure OpenAI$10$30Serverless
GPT-4 Turbo PreviewAzure OpenAI$10$30Serverless
GPT-4 Vision PreviewAzure OpenAI$10$40Serverless
GPT-4 TurboOpenRouter$10$30Serverless
GPT-4 Turbo PreviewOpenRouter$10$30Serverless
GPT-4 TurboVercel AI Gateway$10$30Serverless
GPT-4OpenAI API$30$60Serverless
GPT-4Azure OpenAI$30$60Serverless
GPT-4OpenRouter$30$60Serverless

Popular comparisons in this family