Don't touch it!
retained text
I know it's unhealthy to have any attachment to an LLM model, and it's pathetic begging a multi-billion-dollar company, but please, Anthropic, don't press a button, pull a lever, or turn a knob that could affect the current performance of Opus 5.5. By all means, release your new Sonnets, Fables, Anecdotes (I kinda imagine that's what Anthropic will call their JEV-like decision model) and even Opus models. They will have their use cases and might be better and more useful, but if they are not, we will always have Opus 5.5 to come home to, apologize for our misstep, and continue working. We will optimize our harnesses around it with skills, hooks, and other models. As long as the base doesn't change, we will be completely fine, and the current capabilities are more than enough for most people. Past experiences with Anthropic tell us that this won't last, and the model will get nerfed or dumbed down to free up compute. But one can hope that this time it's different, right? Right?
document fetched 2026-09-27 00:23 · canonical https://reddit.com/r/ClaudeAI/comments/1wqjk4a/dont_touch_it