The quiet crisis in prompt caching
The short version is in the deck. The long version is in the tickets.
The conversation around prompt caching keeps returning to the same trade-off, usually because nobody wants to price it.
Subscriber story
Continue reading with a subscription
This piece is part of the subscriber archive. Subscribe to the daily brief for full access — one email, no card, cancel any time.
Already a subscriber?
#training#infrastructure#inference
More in this section
🔒aiSubscriber
Why small models are winning on-device
Four-bit quantization plus better distillation is making 3B models genuinely useful in production.
🔒aiSubscriber
Synthetic data is hitting a wall nobody wants to name
Model collapse is real, but the more immediate problem is boring: attribution and licensing.
ai
The open-weights licence split is becoming permanent
Two families of licence, two incompatible expectations, and a lot of downstream confusion.