@TylerAlterman there's a monologue in the backrooms webseries that I think sums it up well. in six hundred million million square miles of the complex, our world is the anomaly.
@DosMenoncin@FactoryAI@AmpCode factory is better than amp because factory has pretty good discounts on in house models, while amp is sort of gimmicky and needs api billing rates
using gpt 5.5 pro is a weird vibe. i've felt stuff like this before (first time using the openai playground in 2023, using sonnet 4.5 in cursor) but this is the first time ever I feel like I'm communing with some kind of great intellect obelisk of humanity rather than a really good text predictor.
Minimax M3 results are now live on GBENCH:
It's a solid model, but the other Chinese labs with April releases had slightly better models.
The main thing to worry about is benchmaxxing -- their model card was NOT accurate.
Our evaluations are designed to resist this kind of overfitting.