@macrocephalopod@__paleologo The fundamental assumption in this is equal weight, right? As periodicity can also be “appropriately” captured by half-life?
@thdxr It’s a very weird model, super fast which is great to work with, in the other hand it feels a whole high rise below the bench marks most of the time.
@jetfam@thsottiaux LAWSUIT! This is illegal in many jurisdictions and there are already lawsuits against Anthropic in the US. EU will follow and has even stricter laws against unilateral changes of terms.
A ChatGPT Pro routing bug may be worse than it looked.
One affected user says GPT-5.6 Pro can fall back to GPT-5.5 mini for ~50 hours after it triggers.
Multiple Pro users have reported the same model mismatch.
OpenAI still hasn’t publicly explained it.
@AnthropicAI@alexalbert__ “Our watermarking method doesn’t have any practical impact on the quality or content”
This is a lie.
You literally say you change the wording
Anyone with common sense about how psychology, writing, and legal documents work, knows this is completely false