@TokenGremlin I think this will flow onto all other plans as well unfortunately, and then the $500 plan will have the same usage as the previous 20x. And I don't trust Ant to keep their limits for long either. Terrible time for AI
@meghrajai they have to, otherwise Ant is going to capatilise even more and it will be even more brutal for openai. It is already pretty crappy for them with the limits cut.
@sgdd666@eidzoku yep my experience with gpt vs claude models as well. Claude has consistently just been exceptional at frontend and ui taste so it's much easier to get a good first impression of those models than for backend stuff that astra does better in.
@TokenGremlin OpenAI is by far the best in this regard, especially peopl like Eric @pvncher . Very open minded guy and a great asset for their image on here.
@Ore886@pvncher@TokenGremlin yep 6 astra is very nice as is 5.6 sol. 5.6 luna is also beyond exceptional for it's weight class, I use that almost exclusively at my work because we don't have access to 5.6 sol or astra.
@pvncher@TokenGremlin Thanks for that man, for me at least, that means so much more than any model release. Geneuinly appreciate you listening, more than I can say on a reply here. Gonna keep on rooting for ya🤝
@bornio@pvncher Man I hate whenever I see the word "distinction". Such an overused canned phrase, idk why 6sol does it so often. This is from a medium sized thread. @pvncher
@pvncher@TokenGremlin 6 sol has been exceptional in terms of usage limits and efficiency, I am not hating on the model all round. But it just lacks taste and understanding unfortunately. Happy to send you some short threads as examples, but I don’t wanna spam your inbox
@pvncher@TokenGremlin Hey Eric, I really do love the work on codex you’ve been doing and I genuinely wanna see you guys win. 6sol has just missed the mark a little bit for us, especially in communicating and understanding user intent behind questions. I really hope you can take our feedback to heart
@TokenGremlin I am so sad seeing these responses lately. I want to root for OpenAI because their employyes are genuinely nice, and belive I will continue rooting for them, but I feel like their responses to 6sols poor output has been a little dismissive. Very out of character I hope I’m wrong
@JordiVonB@Im_IrushiK You get really good usage with Sol, but I haven't honestly compared it to Astra low/medium yet. I would assume you get a little bit better usage and performance via sol max than Astra on low or medium. 👍
@JordiVonB@Im_IrushiK and get Astra to review and verify those findings or upgrade the plan of attack. Then get sol max to do the implementation. Then switch back to that Astra thread and get it to review the implementation. So I use astra as the verifyer/planner and sol as the implementer