review of gpt 5.2 after a day of hard use - sometimes really slow but good at most times - does what its told so its at par with opus 4.5 - opus 4.5 is still better at code (bias towards speed) - grabs context from codebase pretty well - ui still has that slop look by default heard that there’s more to come so will wait a bit more overall a great general purpose model but not the best at times for coding

code can be written by llms dev ops should be next why is there no agent that’s cooking with aws or k8s?

After playing with Cursor for hours after the announcement I can confirm that GPT-5.1 Codex Max Extra High Fast is the best coding model right now with Opus 4.5 close second