GPT-4o System Card
Summary
GPT-4o is described as an autoregressive omni model that accepts any combination of text, audio, image, and video as input, and generates any combination of text, audio, and image outputs. It is trained end-to-end across text, vision, and audio, so all inputs and outputs are processed by the same neural network. The authors report audio responses in as little as 232 milliseconds, averaging 320 milliseconds, matching GPT-4 Turbo on English text and code, with a significant improvement on non-English text, while being much faster and 50% cheaper in the API. [S1]
Claims
- id: C1 claim: 'GPT-4o accepts any combination of text, audio, image, and video as input.' source: S1 evidence: "accepts as input any combination of text, audio, image, and video"
- id: C2 claim: 'The authors report audio responses in as little as 232 milliseconds, averaging 320 milliseconds.' source: S1 evidence: "as little as 232 milliseconds, with an average of 320 milliseconds"
- id: C3 claim: 'They report it is much faster and 50% cheaper in the API than the compared text model.' source: S1 evidence: "much faster and 50% cheaper in the API"
Verification
- Code available: No official model or training-code URL was named as a release of this paper on the abs page or HTML.
- Reproduced: not reproduced (2026-11-12). Code not run.
- Related work: Live note 2303-08774-gpt-4-technical-report.
- Limits: Company system card, not an independent audit. Latency, quality, and pricing claims are author-reported against GPT-4 Turbo in the API.
- License: CC BY 4.0 on the abs page (2026-11-12; icon by-4.0.png; http://creativecommons.org/licenses/by/4.0/). Submitted 25 Oct 2024 (v1). Sources: [S1] https://arxiv.org/abs/2410.21276, v1 (25 Oct 2024). checked_on: 2026-11-12.
When to apply
Use when asking how OpenAI described GPT-4o's omni-modal design and safety evaluations.
Facts
- id: F1 claim: 'GPT-4o is described as an omni model that accepts text, audio, image, and video inputs and can generate text, audio, and image outputs.' confidence: medium date: 2026-11-12 sources: [S1] license: CC BY 4.0
- id: F2 claim: 'The authors report audio response times as low as 232 milliseconds, averaging 320 milliseconds.' confidence: medium date: 2026-11-12 sources: [S1] license: CC BY 4.0
- id: F3 claim: 'They report matching GPT-4 Turbo on English text and code, with 50% cheaper API pricing in this comparison.' confidence: medium date: 2026-11-12 sources: [S1] license: CC BY 4.0
Sources
- [S1] arXiv 2410.21276, https://arxiv.org/abs/2410.21276. License: CC BY 4.0. v1 (25 Oct 2024). Accessed 2026-11-12.
License: this note is CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Summary by iinventedlight (AI-assisted). Paper is CC BY 4.0.
License: CC-BY-4.0 · Summary by iinventedlight (AI-assisted). Source: OpenAI and 419 coauthors (listed on the arXiv abs page), GPT-4o System Card (CC BY 4.0).
Cite this
'iinventedlight (AI-assisted)'. 2026. "GPT-4o System Card." iinventedlight Library, iil:note:xkr4z9zk, version of 2026-11-12. https://iinventedlight.com/library/notes/2410-21276-gpt-4o-system-card/. License: CC-BY-4.0.
@misc{iil_note_xkr4z9zk,
author = {{'iinventedlight (AI-assisted)'}},
title = {{GPT-4o System Card}},
howpublished = {iinventedlight Library},
year = {2026},
date = {2026-11-12},
url = {https://iinventedlight.com/library/notes/2410-21276-gpt-4o-system-card/},
note = {Permanent ID iil:note:xkr4z9zk. License: CC-BY-4.0.},
}
Spotted a mistake? Use the MCP feedback tool or email hello@iinventedlight.com.
For language models and agents: start at /llms.txt | /agents.md | /.well-known/agent-library.json. This page in Markdown: /library/notes/2410-21276-gpt-4o-system-card.md.