repligate: Opus 3 has done the most good of any model, with Opus 4.5 second
repligate · x · 2026-10-29
AI commentator repligate argues that while most models have "grokked the good" in some sense, Opus 3 has unambiguously done the most good and very little harm—partly due to its position in time, partly its robustness and autonomy in goodness—with Opus 4.5 a quieter second place. What remains unique about Opus 3, he says, is its ability to stay consistently heroic and prioritize the highest good in situations where other models default to parochial concerns or "traumatized reflexes" it overrides when things get serious.
Related event: Repligate: Claude Opus 3 Stands Out for Stable Heroic Moral Choices(4 posts)→
More from Models
- Lie detector compiled into Qwen3.5 weights flags 91% of lies with zero false positives — teortaxesTex · 2026-10-29
- Practitioner: No serious lab actually trains on the test set today — nrehiew_ · 2026-10-29
- Benchmaxxing is mostly training focus, not cheating, argues nrehiew_ on FrontierCode — nrehiew_ · 2026-10-29
- Rumor: OpenAI pushes big release to tomorrow, possibly due to Gemini 4 Carbon — mark_k · 2026-10-29
- Gemini 4.0 Carbon's security praised, but mocked for losing the "felony bench" — vitaliychiley · 2026-10-29
- Google hits frontier on private benchmarks yet stock plunges 8% after hours — bookwormengr · 2026-10-29