GPT-4o: Difference between revisions
JavaCoder20 (talk | contribs) →Applications: Removed "Applications" section as it had no citations and felt like AI generated. Tags: section blanking Mobile edit Mobile app edit Android app edit |
→Capabilities: Rewrote sentence for better flow, wording and ease for reader. Added missing punctuation. |
||
| Line 24: | Line 24: | ||
== Capabilities == |
== Capabilities == |
||
GPT-4o achieves state-of-the-art results in voice, multilingual, and vision benchmarks, setting new records in audio speech recognition and translation.<ref name="Hello GPT-4o">{{Cite web |title=Hello GPT-4o |url=https://openai.com/index/hello-gpt-4o/ |website=OpenAI}}</ref> GPT-4o scores 88.7 on the Massive Multitask Language Understanding ([[MMLU]]) benchmark compared to 86.5 by GPT-4.<ref name="Hello GPT-4o" /> for Voice-to-Voice, unlike GPT-3.5 and GPT-4 which convert the voice to text, give the text to the model then convert the text back to voice using another model, GPT-4o natively supports Voice-to-Voice making the response near instant and seamless<ref name="Hello GPT-4o" /> |
GPT-4o achieves state-of-the-art results in voice, multilingual, and vision benchmarks, setting new records in audio speech recognition and translation.<ref name="Hello GPT-4o">{{Cite web |title=Hello GPT-4o |url=https://openai.com/index/hello-gpt-4o/ |website=OpenAI}}</ref> GPT-4o scores 88.7 on the Massive Multitask Language Understanding ([[MMLU]]) benchmark compared to 86.5 by GPT-4.<ref name="Hello GPT-4o" /> for Voice-to-Voice, unlike GPT-3.5 and GPT-4 which convert the voice to text, give the text to the model then convert the text back to voice using another model, GPT-4o natively supports Voice-to-Voice making the response near instant and seamless.<ref name="Hello GPT-4o" /> |
||
The model supports over 50 languages,<ref name="TechCrunch" /> covering over 97% of speakers. Mira Murati demonstrated the model's multilingual capability by speaking Italian to the model and having it translate between English and Italian during the live-streamed OpenAI demo event on 13 May 2024. In addition to this, the new tokenizer uses fewer tokens for certain |
The model supports over 50 languages,<ref name="TechCrunch" /> covering over 97% of speakers. Mira Murati demonstrated the model's multilingual capability by speaking Italian to the model and having it translate between English and Italian during the live-streamed OpenAI demo event on 13 May 2024. In addition to this, the new tokenizer uses fewer tokens for certain languages, especially non-English languages such as Gujarati which uses 4.4x fewer tokens, making it cheaper for those languages.<ref name="Hello GPT-4o" /> |
||
It is currently the leading model in the Large Model Systems Organization (LMSYS) [[Elo rating system|Elo]] Arena Benchmarks by the [[University of California, Berkeley]].<ref>{{Cite web |last=Fedus |first=William |title=GPT-4o is our new state-of-the-art frontier model. |url=https://twitter.com/LiamFedus/status/1790064963966370209}}</ref> |
It is currently the leading model in the Large Model Systems Organization (LMSYS) [[Elo rating system|Elo]] Arena Benchmarks by the [[University of California, Berkeley]].<ref>{{Cite web |last=Fedus |first=William |title=GPT-4o is our new state-of-the-art frontier model. |url=https://twitter.com/LiamFedus/status/1790064963966370209}}</ref> |
||
Revision as of 15:20, 15 May 2024
 | This article contains promotional content. (May 2024) |
| Developer(s) | OpenAI |
|---|---|
| Initial release | May 13, 2024 |
| Predecessor | GPT-4 Turbo |
| Type | |
| License | Proprietary |
| Website | openai |
GPT-4o (GPT-4 omni) is a multilingual, multimodal generative pre-trained transformer designed by OpenAI. It was announced by OpenAI's CTO Mira Murati during a live-streamed demo on 13 May 2024 and released the same day.[1] GPT-4o is free, but with a usage limit that is 5 times higher for ChatGPT Plus subscribers.[2] Its API is twice as fast and half the price of its predecessor, GPT-4 Turbo.[1]
Background
GPT-4o was originally shadow launched on LMSYS, as 3 different models. These 3 models were called gpt2-chatbot, im-a-good-gpt2-chatbot, and im-also-a-good-gpt2-chatbot. On 7 May 2024, Sam Altman revealed that OpenAI was responsible for these mysterious new models.[3]
Capabilities
GPT-4o achieves state-of-the-art results in voice, multilingual, and vision benchmarks, setting new records in audio speech recognition and translation.[4] GPT-4o scores 88.7 on the Massive Multitask Language Understanding (MMLU) benchmark compared to 86.5 by GPT-4.[4] for Voice-to-Voice, unlike GPT-3.5 and GPT-4 which convert the voice to text, give the text to the model then convert the text back to voice using another model, GPT-4o natively supports Voice-to-Voice making the response near instant and seamless.[4]
The model supports over 50 languages,[1] covering over 97% of speakers. Mira Murati demonstrated the model's multilingual capability by speaking Italian to the model and having it translate between English and Italian during the live-streamed OpenAI demo event on 13 May 2024. In addition to this, the new tokenizer uses fewer tokens for certain languages, especially non-English languages such as Gujarati which uses 4.4x fewer tokens, making it cheaper for those languages.[4]
It is currently the leading model in the Large Model Systems Organization (LMSYS) Elo Arena Benchmarks by the University of California, Berkeley.[5]
See also
References
- ^ a b c Wiggers, Kyle (2024-05-13). "OpenAI debuts GPT-4o 'omni' model now powering ChatGPT". TechCrunch. Retrieved 2024-05-13.
- ^ Field, Hayden (2024-05-13). "OpenAI launches new AI model GPT-4o and desktop version of ChatGPT". CNBC. Retrieved 2024-05-14.
- ^ Sam Altman "https://twitter.com/sama/status/1787222050589028528" Twitter, X. Retrieved 14 May 2024.
- ^ a b c d "Hello GPT-4o". OpenAI.
- ^ Fedus, William. "GPT-4o is our new state-of-the-art frontier model".
| Products | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| Foundation models | |||||||||
| People |
| ||||||||
| Related | |||||||||
