Opus 4.6 Performance was way better this morning. Between 10 AM and noon I was able to get Opus 4.6 to generate improvements to my employer's SaaS tool that will reduce our monthly cloud spend by 20-25%.
Since 12 PM noon they've scaled back the Opus 4.6 to sub-GPT-4o performance levels to cheap out on query cost. Now I can barely get this thing to generate a functional line of python.
Since 12 PM noon they've scaled back the Opus 4.6 to sub-GPT-4o performance levels to cheap out on query cost. Now I can barely get this thing to generate a functional line of python.