π11π₯4
Bet it didnβt have many sideways heads in the training data, wants to put them vertical
π³πΎπΎπΌπΏπΎπ π πΈπ½πΆ
π³πΎπΎπΌπΏπΎπ π πΈπ½πΆ
π5
Finally a true instruct model for images
Been saying something like this is needed for images for years
π³πΎπΎπΌπΏπΎπ π πΈπ½πΆ
Been saying something like this is needed for images for years
π³πΎπΎπΌπΏπΎπ π πΈπ½πΆ
π₯5π«‘5β€βπ₯1
GPT-4o API unfortunately doesnβt seem to support the new image modality yet
Would bet that this new update isnβt even GPT-4o at all, but rather a new model that GPT-4o calls as a tool
Looking into it
π³πΎπΎπΌπΏπΎπ π πΈπ½πΆ
Would bet that this new update isnβt even GPT-4o at all, but rather a new model that GPT-4o calls as a tool
Looking into it
π³πΎπΎπΌπΏπΎπ π πΈπ½πΆ
π2