Maybe Sam Altman’s strawberry image was a secret message in spite of everything, as roughly per week after the OpenAI CEO posted it on social media the corporate has confirmed that ChatGPT acquired a shock replace final week that it hadn’t informed anybody about.
Making this recognized on social media, an official ChatGPT account revealed that “there is a new GPT-4o mannequin out in ChatGPT,” including in a follow-up that this is not a brand new mannequin like a GPT-5 improve, simply an enchancment to the prevailing GPT-4o mannequin.
there is a new GPT-4o mannequin out in ChatGPT since final week. hope you all are having fun with it and test it out if you have not! we predict you will prefer it 😃August 12, 2024
In its Mannequin Launch Notes we’re given a imprecise description of what these upgrades embrace – specifically “Bug fixes and efficiency enhancements” based mostly on “experiment outcomes and qualitative suggestions” – which isn’t a lot to go on.
OpenAI admits that there’s not a brand new functionality or a particular improve it will possibly level to, even calling itself out for not having a quantitative benchmark which it will possibly use to elucidate these smaller AI conduct tweaks. So we simply need to belief it, and have a look at anecdotal experiences that GPT-4o now works higher than it did earlier than.
It’s additionally unclear if this enchancment is the ‘Strawberry’ undertaking we’ve been ready for. This can be a codename which leaked earlier this 12 months for the following ChatGPT mannequin replace, and whereas this newest enchancment would rely as an replace, and should have been what Sam Altman was teasing with that strawberry picture, leakers have been anticipating one thing a bit extra substantial for Strawberry.
It does, nevertheless, elevate the query of the way you observe AI enhancements. You’ll be able to talk about the quantity of knowledge an AI has entry to, or the processing energy of the chipsets powering AI on gadgets, however what the top person cares about is how helpful the AI really is, which is a tough factor to quantify because it’s based mostly on individuals’s notion, relatively than one thing you can measure. Here is hoping somebody (OpenAI or in any other case) comes up with an efficient benchmarking instrument quickly.
Regardless, if you wish to really use the brand new ChatGPT mannequin we have now excellent news, it has already changed the outdated model of GPT-4o contained in the instrument. So merely go to ChatGPT and also you’ll have the ability to discuss to the brand new bot identical to you probably did the final model of GPT-4o – although without cost customers it’ll include a message restrict.
Alternatively, you may wish to take a look at the entire Gemini AI upgrades Google introduced for its upcoming Google Pixel 9 telephones. Options like Gemini Reside and Add Me appear very spectacular already, and we anticipate the instruments will solely get higher.
You may also like
