Cactus Compute Cactus develops compact automation foundation models that run locally on devices, with models sized from 8–29 MB, 2-bit quantization, and up to 4,000 tokens per second of decode speed.
Cactus Compute Cactus develops compact automation foundation models that run locally on devices, with models sized from 8–29 MB, 2-bit quantization, and up to 4,000 tokens per second of decode speed.