Today's the day! OpenAI, the pioneering AI research lab, has unleashed its most advanced text-generating model, GPT-4, making it generally accessible through its API. If you're an existing OpenAI API developer with a track record of successful payments, then the future is now!
The anticipation has been immense. Since its reveal in March, millions of developers have lined up for access to the GPT-4 API. It's not just the ability to generate human-level text, including code, that's excited them. It's the added capability of accepting both image and text inputs, a noteworthy step up from GPT-3.5, its predecessor, which only accepted text.
While GPT-4's image understanding ability isn't accessible to all OpenAI customers just yet, OpenAI is testing it with an exclusive partner, Be My Eyes. There's no word yet on when it will open up to the wider customer base, but the excitement is palpable.
It's fair to say, though, that GPT-4 isn't perfect. Like most generative AI models today, it sometimes "hallucinates" facts and makes reasoning errors with confidence. And, it doesn't learn from its experience, sometimes causing hiccups like introducing security vulnerabilities into the code it generates.
Looking ahead, OpenAI plans to allow developers to fine-tune both GPT-4 and GPT-3.5 Turbo with their own data, a feature that should roll out later this year.
The generative AI field has become a hotbed of competition since GPT-4's reveal. Recently, Anthropic increased the context window of its AI model, Claude, from 9,000 to 100,000 tokens, outpacing GPT-4's previous high of 32,000 tokens.
OpenAI also making its DALL-E 2 and Whisper APIs generally available. And, starting January 4, 2024, older OpenAI models, specifically GPT-3 and its derivatives, will be phased out, replaced with more efficient base GPT-3 models. OpenAI commits to provide support to users during this transition, ensuring a smooth shift towards the new models.
Get ready for a whole new chapter in AI technology with OpenAI's groundbreaking GPT-4!
