here is GPT-4, our most capable and aligned model yet. it is available today in our API (with a waitlist) and in ChatGPT+.
openai.com/research/gpt-4
it is still flawed, still limited, and it still seems more impressive on first use than it does after you spend more time with it.
Conversation
it is more creative than previous models, it hallucinates significantly less, and it is less biased. it can pass a bar exam and score a 5 on several AP exams. there is a version with a 32k token context.
34
152
1,723
we are previewing visual input for GPT-4; we will need some time to mitigate the safety challenges.
Replying to
we now support a “system” message in the API that allows developers (and soon ChatGPT users) to have significant customization of behavior. if you want an AI that always answers you in the style of Shakespeare, or in json, now you can have that.
22
100
1,300
we are open-sourcing OpenAI Evals, our framework for automated evaluation of AI model performance, to allow anyone to help improve our models.
20
91
1,237
we have had the initial training of GPT-4 done for quite awhile, but it’s taken us a long time and a lot of work to feel ready to release it. we hope you enjoy it and we really appreciate feedback on its shortcomings.
69
52
1,583
Replying to
Did you consider visual outputs as well as inputs? Seems that would help with chain of thought reasoning etc
Replying to
Just signed up with “Be My Eyes”. Is that the best way to test this?
Replying to
Do that because the trolls will test it and post it on twitter about how they made GPT4 do or say something bad lol
2
Replying to
Assistant called you out hard on this 😬 as you’re opening it up too, can you try a new thread on danger?
Replying to
Probability that, eventual, something you build causes human extinction?
1