Post

Did we ever figure out what period that $65B was based on? Was it a full quarter, half year, month?

I think it was almost entirely OpenAI gaining back market share with some good models while everyone was mad at Opus 5 and Fable was too expensive

not a specific one, no I usually test new models on whatever spec I'm currently working on usually nothing huge or complicated

Probably not, but they are probably good enough for lots of firms. Mine had already prohibited Opus usage before exploring open models.

Also, here are 1 quarter and 2 quarter annualized growth rates: Q2 to Q3: (16.8/11.6)^4=4.4 Q1 to Q3: (16.8/4.7)^2=12.8

ramp data (take it with a grain of salt) suggests Opus 5.5 is a winner ramp.com/data/ai-inde...

Ah, I see now. Thx

Not too shocking with the issues they had after Opus 4.8 and before Opus 5.5. But still a bit surprising to see the revenue growth finally slow after such rapid increases.

Github Copilot cut Claude options. If I recall correctly they've recently lost one of their main partners so I'm shocked they've reached that revenue at all, but profitable? Where are they hiding all their losses for Max plans and training?

Interesting! I suspect that when we see actual profit numbers the growth will also have slowed. R&D, training, and inference on much larger models is expensive, especially as models get obsoleted faster.