GPT-4 and Flow XO
GPT-4 has been released! And boy is it powerful! (and expensive!!!)
As of this writing, GPT-4 is still being rolled out based on an invite list. You can request an invitation here.
Once you have been granted access, or GPT-4 is made generally available, you will be able to use it in your Flow XO flows.
If you were accepted into the beta, and you do NOT see gpt-4 as an option, then you probably need to generate a new API key on the Open AI website. We were not able to see GPT-4 as an available model using our API keys generated prior to our being granted access.
GPT-4 is very, very impressive. Our internal testing shows a huge improvement in capability, quality, and reasoning power over GPT-3 and GPT-3.5 Turbo (ChatGPT).
However, it is not cheap. At all. And it can be slower than ChatGPT (which is already slow enough). So once you gain access, and you start playing around, you will find it accomplishes many tasks better than the earlier models - but because of the price, you may want to reserve it for places (like complex reasoning tasks or extensive text generation) where you simply can't make ChatGPT perform well enough. Or, if you have lower chat volume, you may just want to use it in your regular flows, but just be aware you can rack up your costs much more quickly than ChatGPT or even GPT-3.
NOTE: The AP interface for GPT-4 is the same as ChatGPT. Therefore, you can ONLY use GPT-4 in the "Complete Chat" tasks, and tasks derived from Complete Chat, such as Detect User Intent, Translation, and Summarization. It will not show up in the "Complete Text", "Edit Text", or "Generate Image" workflows
The Flow XO Team