Yes, this project seems like a misunderstanding of what Bubeck and team were obsering with their unicorn test. GPT-4 was being trained, and checkpoints were provided to them to experiment with. The improvements in the unicorn reflected further training progress.
The models on offer now are frozen(-ish). Per the models[0] page, the non-snapshot model IDs "[w]ill be updated with our latest model iteration". So this project will eventually hit another version of the model, but (a) one image definitely will not be enough to reliably discern a difference and (b) seems like that 'latest model iteration' cadence will be much, lower than a day.
The models on offer now are frozen(-ish). Per the models[0] page, the non-snapshot model IDs "[w]ill be updated with our latest model iteration". So this project will eventually hit another version of the model, but (a) one image definitely will not be enough to reliably discern a difference and (b) seems like that 'latest model iteration' cadence will be much, lower than a day.
[0]: https://platform.openai.com/docs/models