MLX foundation
Built around MLX, Apple's machine-learning framework for Apple silicon. Forge is being developed with that ecosystem as its execution foundation.
Forge · Inference
Memory-efficient inference, built on MLX.
Forge is Foundation's upcoming inference engine, built on Apple's MLX framework. Its goal is to make model execution more memory-efficient and simplify the path from a prepared model to evaluation and export.
What you get
Memory efficiency
Forge's central goal is to reduce the memory overhead of inference, helping make model execution more practical on compatible local hardware.
Part of the model workflow
Inference is one stage in a broader workflow that includes model preparation, training, benchmark evaluation, and export.
How it works
- 01
Prepare the model
Bring the model into the workflow and get it ready for execution.
- 02
Run inference
Use the MLX-based engine to execute the model with memory efficiency as a primary design goal.
- 03
Evaluate the result
Run benchmark evaluations and prepare the model for export.
More useful work from the memory you have.