PRISM ML × PRISM ENGINE
Prism ML's Bonsai Ternary and binary models work on efficient numerical representation. Prism Engine sits downstream on the deployment scale: compiling those models into target-aware execution artifacts for real machines.
Bonsai and Prism Engine do not need to compete for the same layer. Bonsai can explore training, quantization, ternary representations, binary representations, and model-quality tradeoffs. Prism can accept the resulting model artifact and solve the deployment questions that appear when those weights meet a heterogeneous machine.
The Bonsai Ternary and binary model work gave Prism Engine a demanding upstream representation to compile against. That relationship helps validate the architecture’s central boundary: numerical representation can evolve upstream while deployment layout, execution views, scheduling, and evidence remain explicit downstream.
The same Bonsai model family can become different physical artifacts depending on where it will run. The logical model and quality contract remain attributable to Prism ML; the deployment image reflects the machine and execution objective.
Small resident set, bounded context, conservative fallback.
Metal layout, unified memory, local interactive execution.
Capability-driven views, phase placement, explicit handoffs.
Sharding, KV ownership, transfers, coordination, replay.
Training behavior
Quantization strategy
Ternary / binary representation
Calibration and numerical validation
Representation admission
Physical tile layout
Execution views
Scheduling, residency, and receipts
Prism Engine can expose deployment evidence back upstream: tensor sensitivity, layout pressure, target-specific failures, and runtime observations. Prism ML can use that evidence to improve the next model or representation. The result is a feedback loop without collapsing training and deployment into one system.
The meeting sentence
Prism ML and Prism Engine independently built adjacent layers of the same stack: Bonsai makes efficient models; Prism makes those models deployable across heterogeneous hardware.
PRISM ML × PRISM ENGINE
The architecture is strongest when both systems remain excellent at their own layer and meet at an explicit, inspectable artifact boundary.
Read the architecture ↗ Back to Prism Engine ↗