Altman Lays Out OpenAI API Vision, Corrects 'Video' to 'Voice'

On September 23, OpenAI CEO Sam Altman posted a series of tweets outlining the company's API strategy, then personally corrected a line that had been widely misread. The bottom line: he meant 'voice,' not 'video,' so rumors of an imminent return to video generation were not officially confirmed, while the all-price-tier, all-modality API positioning stands.

Confirmed

Altman set the goal for OpenAI's API: the best model at every price point, and the best in every modality including text, code, image and video.
He said the best ideas will come from the developer community, invited developers to build on the API, and teased an upcoming Dev Day.
Altman clarified he originally meant 'VOICE' rather than 'video,' apologizing for disappointing users.
The posts briefly fueled video-model speculation; user kimmonismus joked that misspeaking 'costs a reset, rules are rules,' reflecting ongoing user grumbling about usage and reset rules.

Unconfirmed

An imminent OpenAI video generation model was only an interpretation of the tweets, effectively walked back by Altman's clarification; no official confirmation existed at posting time.

Why it matters

This was OpenAI explicitly defining its API ecosystem ambition—competing across every price tier and modality rather than on single breakthroughs—and the Dev Day tease suggests a wave of developer-focused updates ahead. The correction also serves as a reminder that voice, not video, was the officially intended direction.

2026-09-23 ~ 2026-09-23 · 6 related posts

Primary sources

Sam Altman: OpenAI API aims for best model at every price point and every modality — sama · 2026-09-23 02:34 (UTC+08:00)
Altman clarifies: OpenAI's new release is voice, not video, 'sorry to disappoint' — sama · 2026-09-23 05:38 (UTC+08:00)

[source] Sam Altman: OpenAI API aims for best model at every price point and every modality — sama · 2026-09-23
Sam Altman strongly suggests OpenAI video generation is coming at DevDay — Neurogence · 2026-09-23
[source] Altman clarifies: OpenAI's new release is voice, not video, 'sorry to disappoint' — sama · 2026-09-23
Sam Altman jokes about burning a reset over a voice-not-video typo — kimmonismus · 2026-09-23
Altman signals OpenAI video model as API aims to be best in every modality — Kyrannio · 2026-09-23

1 near-duplicate retellings: patience_cave