Claude's old Opus 4.6 model selector still resolves

Typing /model claude-opus-4-6[1M] still works, according to a 50-upvote reminder post on r/ClaudeAI. The post's author reaches for the older model specifically when Opus 5 or Fable 5 refuse a task or argue about it. Opus 4.6, in the author's account, just does the work.

That's a change in what picking a model means. It used to be about context length, speed, or cost. Now, for at least some Claude users, it includes which model is least likely to push back.

The thread doesn't say how often Opus 5 or Fable 5 refuse tasks compared to Opus 4.6. It also doesn't say whether the difference is real behavior or just perception. Fifty upvotes suggest other people recognized the pattern, not that anyone measured it.

The post doesn't say what happens if claude-opus-4-6[1M] gets pulled from the model list. Whoever's using it to route around a newer model's judgment calls would lose that option. Nothing in the thread suggests a replacement is coming.