AI chatbot ChatGPT, developed by OpenAI, has gained popularity and currently has over 180 million users worldwide. According to foreign netizens, the latest version, GPT-4.5, seems to have quietly gone online.
Table of Contents:
Six-fold Price Increase for GPT?
Official Response from OpenAI
OpenAI Admits Model Getting Dumber, Developing GPT-5
The AI chatbot ChatGPT, which has been popular worldwide, was previously rumored to release GPT-4.5 as early as the end of December. Although this has not been officially confirmed so far, recent posts on the OpenAI subreddit by netizens suggest that GPT-4.5 may have quietly gone online. The post has been deleted, but leaked screenshots from the post showed GPT-4.5 being touted as “OpenAI’s most advanced model,” with enhanced multimodal capabilities for language, audio, visual, 3D, and video data processing.
According to the leaked screenshots, GPT-4.5 has three models:
– GPT-4.5: $0.06 per 1000 tokens input, $0.18 per 1000 tokens output
– GPT-4.5-64k: $0.12 per 1000 tokens input, $0.36 per 1000 tokens output
– GPT-4.5 Audio & Voice Edition: $0.012 per minute input, $0.024 per minute output
Compared to the GPT-4 Turbo model, GPT-4.5 has increased its price by six times, while GPT-4.5-64k has increased its price by twelve times. Netizens believe that the screenshots are likely from OpenAI’s draft page, indicating that they may have been leaked by internal staff, making them highly credible. Many users even speculate that they are currently using GPT-4.5.
Some users have shared that when they asked ChatGPT about the exact model name of their API response, ChatGPT replied as “gpt-4.5-turbo.” However, OpenAI researcher Will DePue debunked the gpt-4.5 news, calling it a “very weird and strange illusion.” Some netizens analyzed that this could be related to the contamination caused by the prompt “Not ‘ChatGPT with x’,” and people were hoping ChatGPT would answer “gpt-4.5-turbo” as the response.
In summary, the leaked information about GPT-4.5 has not been officially confirmed by OpenAI. However, when netizens directly asked OpenAI CEO Sam Altman about the authenticity of the GPT-4.5 leak, his informal denial of “nah” sparked endless speculation. Previously, many users of ChatGPT complained about GPT-4 becoming less intelligent. OpenAI even acknowledged that GPT-4 had indeed become lazier, stating, “We haven’t updated the model since November 11. It’s not intentional laziness. The model’s behavior is unpredictable, and we are currently looking for solutions.”
On the other hand, OpenAI announced on November 15 that due to overwhelming usage, they decided to temporarily suspend new user registrations for ChatGPT Plus. After nearly a month, they finally resumed new user registrations for ChatGPT Plus. However, some individuals are concerned that reopening registrations may lead to more cases of crashes or poor performance. If user usage increases again in the future, the issue of GPT-4 becoming lazier and less intelligent may become even more difficult to solve.
To address these problems, a more efficient generation model may be needed. Sam Altman revealed in November that they are developing the next-generation model, GPT-5, but he did not promise a release date. He only mentioned that this new model would require more data for training, including publicly available data from the internet and proprietary data provided by companies.
Related Reports:
– EU Passes Major AI Legislation! ChatGPT, Large Language Models, Facial Recognition… All to be Regulated
– Why is Google’s New AI Model Gemini Powerful? iKala Founder: ChatGPT Can’t Compete with Google’s Ecosystem
– Crushing GPT-4! Google Unveils “Gemini Native Multimodal Model”: AI Surpasses Human Understanding, Offline Usability, Equipped on Pixel 8 Pro
Tags:
AI
ChatGPT
GPT-4.5
OpenAI