Here’s everything we know about GPT-4.5: - The model will be named GPT-4.5o, as numerous leaks suggest it will be a continuation of the multimodal GPT-4o. Multimodal means integrating multiple types of data, such as text, images, video and audio. This means GPT-4.5o should feature an improved version of GPT-4o’s advanced voice mode. The previous version was criticized for being unclear and inaccurate in its responses. 🧵

串文

2024-10-17 19:25

讚

回覆

轉發

作者

Mikumice
mikumice

粉絲

串文

69+

讚

回覆

轉發

24小時粉絲增長

無資料

互動率

(讚 + 回覆 + 轉發) / 粉絲數

53.33%

回覆 (BETA)

最先回覆的內容
發文後	用戶	內容
幾秒內	Mikumice mikumice	- It is estimated to be 4 to 5 times larger than its predecessor in terms of the number of parameters. - Its effective performance, through synthetic data in its training, is expected to be even more than its increase in parameter size - The model has undergone red teaming (a process where experts attempt to find vulnerabilities or flaws by acting as adversaries) and is currently awaiting release. - GPT-4.5o is positioned as a competitor to Claude Opus 3.5.