Tiiny

Tiiny Runs a 10B-Plus Model in Your Palm, but Local AI Still Has a Speed Problem

Tiiny Pocket runs local AI in a 300g device at up to 64 tokens per second, but larger models remain far slower than cloud services.