AI Radar

GPT-4o

Available · Language, Multimodal · Milestone

GPT-4o is an OpenAI model released on 13 May 2024 in ChatGPT, including the free tier, and in the API. OpenAI describes it as an omni model trained end to end across text, vision and audio; at launch it offered text and image input with text output, and audio followed later, for example in ChatGPT Advanced Voice from September 2024. [1] [2] Primary source

Timeline of GPT-4o →

Claims and evidence

  • Released Text and image capabilities began rolling out in ChatGPT and the API on this day; audio and video capabilities were to follow. Primary source[1]Model availability [3]May 13, 2024
  • Snapshot gpt-4o-2024-08-06 released with Structured Outputs Primary source[4]Introducing Structured Outputs in the API [3]Aug 6, 2024
  • Advanced Voice using GPT-4o native audio rolled out in ChatGPT Primary source[2]Advanced voice (September 24, 2024)
  • Snapshot gpt-4o-2024-11-20 released Primary source[3]Nov 20, 2024
  • Retired from ChatGPT Primary source[2]Retiring GPT-4o and other legacy models (February 13, 2026)
  • Status AvailableAPI status on 2026-10-01: the gpt-4o alias and the snapshots gpt-4o-2024-08-06 and gpt-4o-2024-11-20 are not on the deprecations page; gpt-4o-2024-05-13 is deprecated with shutdown on October 23, 2026, and chatgpt-4o-latest was shut down on 2026-02-17. GPT-4o was retired from ChatGPT on 2026-02-13. Primary source[5]Model details; Snapshots [6]2026-04-22: Legacy GPT model snapshots; 2025-11-18: chatgpt-4o-latest snapshot [2]Retiring GPT-4o and other legacy models (February 13, 2026)
  • Successor of GPT-4 TurboThe announcement compares GPT-4o with GPT-4 Turbo throughout; the GPT-4 Turbo model page recommends a newer model like GPT-4o. Primary source[1]first paragraph; Model availability [7]description
  • Change · Modality OpenAI says it processes audio natively with text and images in one model trained end to end, where the earlier Voice Mode chained three separate models.Compared with GPT-4 Turbo Primary source[1]
  • Change · Efficiency In the API it is twice as fast and half the price of GPT-4 Turbo, with five times higher rate limits.Compared with GPT-4 Turbo Primary source[1]
  • Change · Languages A new tokenizer needs fewer tokens for many non-English languages; OpenAI shows the biggest reductions for Indian languages such as Gujarati, Telugu and Tamil.Compared with GPT-4 Turbo Primary source[1]
  • Change · Availability Offered in ChatGPT to free-tier users as well as paid plans.Compared with GPT-4 Turbo Primary source[1]
  • Input text, imageReleased modalities of the gpt-4o API model. OpenAI describes GPT-4o as an omni model trained end to end that accepts any combination of text, audio, image and video and can output text, audio and image (announcement, system card); audio reached users later through ChatGPT voice and separate audio and realtime API models. Primary source[5]Model details [1]Model safety and limitations; Model availability [8]Introduction
  • Output text Primary source[5]Model details
  • Feature Function calling Primary source[5]Supported features
  • Feature Fine tuning availableGeneral availability of fine-tuning for gpt-4o-2024-08-06 from 2024-08-20. Primary source[3]Aug 20, 2024 [5]Supported features
  • Context window 128,000 tokens Primary source[5]Model details
  • Parameters Not disclosedNot disclosed by OpenAI. No claim madeNo source
  • Access consumer app, API Primary source[1]Model availability

Lineage

Predecessors

Successors

Based on

Not derived from another model.

Variants and derived

None recorded.

Siblings

None recorded.

All ancestors

All descendants

Variants

No variants recorded in this record.

Related AI Radar coverage

All model releases from OpenAI on AI Radar →