Some friends and I built a simple mobile app that lets users query OpenAI’s GPT APIs. So far it’s experienced relative success.
We’d like to spend some more money on marketing it - we did so in the past and it worked well. Since OpenAI released their own mobile app, we’ve put a hold on marketing to see how things played out.
Ultimately, we are well aware that our entire business model hinges on OpenAI’s strategic decisions. Things like ending 3rd-party API access, or changing the API pricing could put an end to things overnight.
Is it just us thinking about this? Do we bite the bullet and absorb the risk like thousands of others are? I did some searching but didn’t find anything notable. Many people say that OpenAI is not that interested in providing the services themselves, rather the capabilities, but this is all speculation.
I don’t think OpenAI will do it overnight. And even if they do, Azure will try push back and still offer AI services for a while longer.
Overall, I think it’s important to stay vendor agnostic, because OpenAI’s been quite fickle in the past. Luckily, most competitors are only about a year behind in tech, and hopefully catching up. If your product relies heavily on GPT-3.5, I wouldn’t be worried. GPT-4 might still be an issue, but the gap is starting to close.
If you’re essentially just reselling the API, you could take the step to integrate alternative options into your product - a Mixtral finetune for example has certain strengths that OpenAI can never close the gap on.
Agreed. OpenAI is my number one source, but it would be wise to diversify just in case.
I’ve not tried to finetune Mistral, but I have tested it’s API, and it’s fantastic. It, and Anthropic Claude, are certainly the top contenders for the backup team in case OpenAI decides to leave us hanging. But, I don’t think they’ll do that. Will they?
