importantSYS.SOURCE: TechCrunch• 2026-09-17T22:34:09Z
PrismML's Bonsai 2 27B: A Compact LLM for On-Device AI Applications
PrismML's Bonsai 2 27B compresses a 27B open-source model to 5.9GB while maintaining 98% of original benchmark performance, enabling on-device AI applications. The startup's ternary weight compression technique allows smaller models to run on consumer devices without significant accuracy loss.
If AI lab PrismML isn't on your radar yet, it should be.
*** END OF TRANSMISSION ***