Home | Notifications | New Note | Local | Federated | Search | Logout
Note Detail
7666@7666@comp.lain.la (2026-09-29 10:33:37)
i have been saying this for a long ass time. opus 4.8 class models are completely sufficient for paired programming and automated functions for 99.99% of use cases and these new models only benefit thinking on extremely complicated (like PhD tier) problems.
the capabilities of most last gen frontier models were enough to basically accelerate and assist me with just about any task I felt comfortable offloading *with me still being in the driver's seat of what i am trying to do*.
i guess these newer models are really only showing meaningful gains for losers who just slop out 40 billion lines of code a day and deal with pure vibe coded architectures that they don't understand and that are bloated to all hell.
---Attachments---
image: https://comp.lain.la/media/5e/04/43/5e04438ab499b794b4aad8e4236c5b4d91efec3e9f8287f88d10cba191425bf0.webp
---Reply---
Blurry Moon@sun@shitposter.world (2026-09-29 10:35:39)
@7666 I use chatgpt terra for basic stuff and bump it up to sol for things that can't be fucked up, then Astra now for problems where I don't totally even understand the data or problem yet, to help me figure it out.
Reply
---Replies---
Blurry Moon@sun@shitposter.world (2026-09-29 10:35:57)
@7666 luna is good for questions and shell scripts, it fucks up everything though