|
|
|
|
|
|
by minimaxir
131 days ago
|
|
|
You are selection biasing towards the most extreme cases of AI absurdity. > GPT-5.3-Codex and Opus 2.6 were released. Reviewers note they're struggling to find tasks the previous versions couldn't handle. The improvements are incremental at best. I have not seen any claims of this other than Opus 4.6 being weirdly token-hungry. |
|
|