I asked Borris at an Anthropic event in SF this week why Opus 5 and Fable 5 seem to forget so much when I give it rapid fire tasks when I'm reviewing a UI or something. He told me to run in safe mode, which didn't help at all.

My tinfoil hat theory is Anthropic is trying to get their new models to take on higher-level longer-running tasks, which has a trade-off against rapid-fire tactical use of an LLM.

For these reasons, I've always found the 5 series models from Anthropic aren't great and use 4.8 for a lot of my work.