Back to the episode map

Research Note

GPT-4o Launch and Retirement State Record

OpenAI announced GPT-4o on May 13, 2024 as an omni model accepting combinations of text, audio, image, and video input and producing text, audio, and image output. The la

Aug 4, 20262 min readBy Dalton Anderson

GPT-4o Launch and Retirement State Record

Launch layer

OpenAI announced GPT-4o on May 13, 2024 as an omni model accepting combinations of text, audio, image, and video input and producing text, audio, and image output. The launch demonstrations, product rollout, and later safety evidence arrived on different timelines.

The preserved E018 outline planned a live conversational demo. It does not establish that the recording used the fully released native speech-to-speech GPT-4o experience. Source-era ChatGPT voice and demonstrated future capabilities must remain separate.

Safety evidence

OpenAI's GPT-4o System Card documents model preparation and evaluated audio risks, including unauthorized voice generation, speaker identification, sensitive-trait attribution, emotional reliance, and audio robustness. The report was published after the E018 recording.

The system card is OpenAI's evaluation record. It does not establish suitability for every user, language, accent, disability, environment, or consequential task.

Sky voice boundary

OpenAI publicly stated that Sky used another professional actor, was cast before outreach to Scarlett Johansson, and was paused after concerns. That is OpenAI's account. The reviewed sources do not establish that Sky was trained on Johansson's voice without permission, and this package should not make that claim.

Similarity, consent, publicity rights, deception, casting, and model-training provenance are different questions.

Current model state

As of July 28, 2026, OpenAI's API model record still documents GPT-4o for API use. OpenAI announced that ordinary ChatGPT access to the GPT-4o text model was retired in February 2026. Voice product naming and underlying models are separate from the retired text-model picker state.

The public record should not say GPT-4o remains the default ChatGPT model or that every current voice session is the same May 2024 product.

Publication boundary

The launch record can preserve May 2024 chronology, demonstrations, staged rollout, system-card evidence, Sky dispute boundary, and current API record. Publication-day checks must reverify current availability and model status.

Sources

Follow the evidence.

  1. June 2024 Recall updateblogs.windows.com
  2. Current Recall privacy and controlsupport.microsoft.com
  3. Current GPT-4o API documentationdevelopers.openai.com
  4. Manage Recall for Windows clientslearn.microsoft.com
  5. Recall security and privacy architectureblogs.windows.com
  6. GPT-4o system cardcdn.openai.com
  7. Spotify episodeopen.spotify.com
  8. Current Recall use and requirementssupport.microsoft.com
  9. OpenAI API deprecationsdevelopers.openai.com
  10. Introducing Copilot+ PCsblogs.microsoft.com
GPT-4o Launch and Retirement State Record