OpenAI’s CEO, Sam Altman, has announced the launch of GPT-4o, an advanced version of the AI technology that powers ChatGPT. GPT-4o offers improved performance and efficiency and will be available to all users as OpenAI rolls it out globally. The technology will soon be offered to enterprise users.
GPT-4o provides GPT-4-level intelligence but is much faster and improves on its capabilities across text, voice, and vision. The company claims that GPT-4o understands and analyses the images better than any other model now in use. The quality and speed of GPT-4o’s language capabilities are enhanced to make advanced AI more widely available. More than 50 languages are now supported by ChatGPT for sign-up and login, user preferences, and other features. The “o” in GPT-4o stands for Omni. The GPT-4o can receive any combination of text, voice, and image as input and produce any combination of outputs in the same way. Its response time to audio inputs can be as quick as 232 milliseconds, with an average of 320 milliseconds; this is comparable to the average human response time.
As per the company, it matches GPT-4 Turbo performance on text in English and code, with significant improvement on text in non-English languages, while also being much faster and 50% cheaper in the API.
Before GPT-4o, users were able to interact using ChatGPT with an average latency of 2.8 seconds (GPT-3.5) and 5.4 seconds (GPT-4). This was known as Voice Mode. It makes use of three models: GPT-3.5 or GPT-4 reads text, transcribes audio to text, and then transforms text back to audio. GPT-4, on the other hand, is unable to convey emotion, recognise tone, distinguish between many speakers, or perceive background noise.
With GPT-4o, OpenAI claims to have trained a single new model end-to-end for text, vision, and audio, which means that the same neural network handles all inputs and outputs.
“We’ve evaluated GPT-4o according to our Preparedness Framework and in line with our voluntary commitments. Our evaluations of cybersecurity, CBRN, persuasion, and model autonomy show that GPT-4o does not score above Medium risk in any of these categories. This assessment involved running a suite of automated and human evaluations throughout the model training process. We tested both pre-safety-mitigation and post-safety-mitigation versions of the model, using custom fine-tuning and prompts, to better elicit model capabilities.”
The GPT-4o is now available to ChatGPT Plus and Team users, and Enterprise users will soon be able to access it as well. In addition to this OpenAI will also roll out to ChatGPT Free with usage limits today. Plus users will have a message limit that is up to 5x greater than free users, and Team and Enterprise users will have even higher limits.
GPT-4o is currently available to developers via the API as a text and vision model.
When using GPT-4o, ChatGPT Free users will now have access to features such as experiencing GPT-4 level intelligence, getting responses(opens in a new window) from both the model and the web, analysing data(opens in a new window) and creating charts, Chat about photos you take, upload files(opens in a new window) for assistance summarising, writing or analysing, discover and use GPTs and the GPT Store and build a more helpful experience with Memory.
Also Read: Google I/O 2024 today: How to watch and what to expect
According to OpenAI, the number of messages that GPT-4o free users are able to send will be limited based on demand and usage. In addition to this, ChatGPT will automatically switch to GPT-3.5 so users can continue their conversations.
OpenAI has also launched a new ChatGPT desktop app for macOS, intended to work seamlessly with anything you do on your computer and is available to both free and premium users. You can ask ChatGPT a question right away by using the keyboard shortcut Option + Space. In addition, users will also be able to snap and share screenshots.
