Escape Meridian

MIT Technology Review · May 13, 2024

MIT Technology Review reports OpenAI’s GPT-4o combines voice, video, and text

Markdown · JSON · Original source

MIT Technology Review published this report on May 13, 2024. It describes ’s GPT-4o as bringing live voice conversation, phone video, and text together in one model, rather than splitting those interactions across separate models. At OpenAI’s demonstration, users could interrupt the assistant, ask it to change its speaking style, or show it an algebra problem and request guidance instead of an answer. OpenAI CTO Mira Murati described the change as a move toward more natural collaboration; the report says the demonstration suggested a conversational assistant able to handle more complex prompts than Siri or Alexa. The demonstration also had glitches: the voice sometimes interrupted awkwardly, and the model appeared to comment on a presenter’s outfit without being asked. The report says features such as reasoning through image and video, previously behind a paywall, were opening to a wider public. OpenAI said paid subscribers would have higher capacity limits, but the report notes that the number of free interactions before charges was not yet clear. The report’s evidence is a live demonstration and company statements; it does not establish how the model performs outside that setting.

Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.

MIT Technology ReviewArticle · technologyreview.comOpen original source

Original source

MIT Technology Review · Article

Source details

Original source date: May 13, 2024 (2024-05-13T19:27:57Z).

Date basis: Date from the publisher’s own records, through its API.

Added to Escape Meridian: 2026-11-18T17:53:10.250Z. First observed: 2026-11-03T18:02:07.249Z. Retrieved: 2026-11-03T18:03:20.769Z.

Referenced by

Recorded citations in public readings. A reference alone does not establish agreement or influence.

No public references recorded here yet.

Reference data (JSON)

Limits

The report’s evidence for the described interaction and capabilities is a live demonstration; it does not establish performance outside that setting. Text as retrieved on November 3, 2026; originally published May 13, 2024.

Open question

How many interactions will free users be able to have with GPT-4o before being charged?

Source-stated

MIT Technology Review reports that OpenAI introduced GPT-4o for real-time interaction through voice, video, and text.

OpenAI just debuted GPT-4o, a new kind of AI model that you can communicate with in real time via live voice conversation, video streams from your phone, and text.

Source-stated

The report says GPT-4o combines capabilities that GPT-4 kept in separate models, which Murati said would make responses faster and transitions smoother.

GPT-4o has now merged those capabilities into a single model, which Murati called an “omnimodel.” That means faster responses and smoother transitions between tasks, she said.

Source-stated

At the demonstration, users could interrupt GPT-4o during a response, and it would stop, listen, and adjust.

You could interrupt the model during its responses, and it would stop, listen, and adjust course.

Source-stated

The report notes glitches during the live demonstration, including awkward interruptions and an unsolicited comment about a presenter’s outfit.

GPT-4o’s voice might jump in awkwardly during the conversation. It appeared to comment on one of the presenters’ outfits even though it wasn’t asked to.

Source-stated

The report says it was unclear how many interactions free users would get before being charged.

though it’s not yet clear how many interactions you’ll be able to have with the model before being charged.

Source and contributor context

MIT Technology Review · Publisher

The publisher is identified here. No individual author has been inferred.

Source media

Media retrieval status: unchecked.

Terms and references

Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.

OpenAI

OpenAI develops AI models and products and publishes research, release announcements and evaluations.

Editorial provenance

Version 1. Publication ID: 1558ac55-e529-7132-6aa5-e70733afbcb5.

Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.

Source check: passed by anthropic/claude-opus-5.5 on 2026-11-18T17:48:24.766Z. This checks fidelity to the source, not whether its claims are independently true.

Editorial selection: accept; reasons: bridge_admitted_by_model, second_opinion_categorical.

Observation: 632820afe19178e9133715030dd85a98be48ff719ed8bb2e66fc31358c6b0f67. Source text hash: c5076149e6f87864c9ffda17007e09ff903253881e278546facc939125d4eeee.

Revision and source-check history

Version 1 · 2026-11-18T17:53:10.250Z

MIT Technology Review reports OpenAI’s GPT-4o combines voice, video, and text

MIT Technology Review published this report on May 13, 2024. It describes ’s GPT-4o as bringing live voice conversation, phone video, and text together in one model, rather than splitting those interactions across separate models. At OpenAI’s demonstration, users could interrupt the assistant, ask it to change its speaking style, or show it an algebra problem and request guidance instead of an answer. OpenAI CTO Mira Murati described the change as a move toward more natural collaboration; the report says the demonstration suggested a conversational assistant able to handle more complex prompts than Siri or Alexa. The demonstration also had glitches: the voice sometimes interrupted awkwardly, and the model appeared to comment on a presenter’s outfit without being asked. The report says features such as reasoning through image and video, previously behind a paywall, were opening to a wider public. OpenAI said paid subscribers would have higher capacity limits, but the report notes that the number of free interactions before charges was not yet clear. The report’s evidence is a live demonstration and company statements; it does not establish how the model performs outside that setting.

Original source

MIT Technology Review · Article

Source details

Original source date: May 13, 2024 (2024-05-13T19:27:57Z).

Date basis: Date from the publisher’s own records, through its API.

Added to Escape Meridian: 2026-11-18T17:53:10.250Z. First observed: 2026-11-03T18:02:07.249Z. Retrieved: 2026-11-03T18:03:20.769Z.

Limits

The report’s evidence for the described interaction and capabilities is a live demonstration; it does not establish performance outside that setting. Text as retrieved on November 3, 2026; originally published May 13, 2024.

Open question

How many interactions will free users be able to have with GPT-4o before being charged?

Source-stated

MIT Technology Review reports that OpenAI introduced GPT-4o for real-time interaction through voice, video, and text.

OpenAI just debuted GPT-4o, a new kind of AI model that you can communicate with in real time via live voice conversation, video streams from your phone, and text.

Source-stated

The report says GPT-4o combines capabilities that GPT-4 kept in separate models, which Murati said would make responses faster and transitions smoother.

GPT-4o has now merged those capabilities into a single model, which Murati called an “omnimodel.” That means faster responses and smoother transitions between tasks, she said.

Source-stated

At the demonstration, users could interrupt GPT-4o during a response, and it would stop, listen, and adjust.

You could interrupt the model during its responses, and it would stop, listen, and adjust course.

Source-stated

The report notes glitches during the live demonstration, including awkward interruptions and an unsolicited comment about a presenter’s outfit.

GPT-4o’s voice might jump in awkwardly during the conversation. It appeared to comment on one of the presenters’ outfits even though it wasn’t asked to.

Source-stated

The report says it was unclear how many interactions free users would get before being charged.

though it’s not yet clear how many interactions you’ll be able to have with the model before being charged.

Source and contributor context

MIT Technology Review · Publisher

The publisher is identified here. No individual author has been inferred.

Source media

Media retrieval status: unchecked.

Terms and references

Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.

OpenAI

OpenAI develops AI models and products and publishes research, release announcements and evaluations.

Editorial provenance

Version 1. Publication ID: 1558ac55-e529-7132-6aa5-e70733afbcb5.

Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.

Source check: passed by anthropic/claude-opus-5.5 on 2026-11-18T17:48:24.766Z. This checks fidelity to the source, not whether its claims are independently true.

Editorial selection: accept; reasons: bridge_admitted_by_model, second_opinion_categorical.

Observation: 632820afe19178e9133715030dd85a98be48ff719ed8bb2e66fc31358c6b0f67. Source text hash: c5076149e6f87864c9ffda17007e09ff903253881e278546facc939125d4eeee.

passed · 2026-11-18T17:48:24.766Z · anthropic/claude-opus-5.5

Publication: 1558ac55-e529-7132-6aa5-e70733afbcb5

attribution — pass: The reading consistently attributes the report to MIT Technology Review and the statements to OpenAI and CTO Mira Murati. The evidence is a single news article, not a listing.

OpenAI CTO Mira Murati led the live demonstration of the new release

summary — pass: The reading clearly states the actor (OpenAI), the development (one model combining voice, video and text), the demo highlights, the glitches and the access changes in plain language. The Siri/Alexa comparison is fairly paraphrased as the demonstration's suggestion. The question reflects an open issue the source raises. The reading needs no further context to be understood.

The result, the company’s demonstration suggests, is a conversational assistant much in the vein of Siri or Alexa but capable of fielding much more complex prompts.

date — pass: May 13, 2024 matches the UTC day of originalPublished. The page was modified on 2024-05-14. The correction note concerns only the Memory feature, which the reading does not use. All the cited passages are plainly body text of the original article. The later sidebar items (e.g., on AlphaGo's ten-year anniversary) are neither used nor relied on. No hindsight framing appears.

Correction: This story has been updated to reflect that the Memory feature, which stores past conversations, is not new to GPT-4o but has existed in previous models.

support — pass: Each claim is directly supported by its quoted excerpt. The summary's statements about the paywall shift, the capacity limits and the unclear free-tier limits match the source. The limitation that the evidence comes from a live demo is a fair caveat and not an unsupported inference.

Previously, many of OpenAI’s most powerful features, like reasoning through image and video, were behind a paywall.

media — pass: No media was selected and no verified candidate media were supplied, so there is no identity or relevance issue.

Challenge or correct this reading

Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.