Local model paging · Mac
Overfit loads only the parts of a compatible mixture-of-experts model needed for each response. The remaining files stay in local storage, so the complete model does not need to fit in memory. This adds capacity, not a guarantee of speed.
Local storage
Model filesverified filesWorking memory
Active model partssized for this deviceApple Silicon
Run the modellocal executionWhy this works
A mixture-of-experts model is made from many specialist parts, but each token uses only a small subset of them. Loading every part into memory is often unnecessary.
Overfit keeps shared parts in memory and stores the rest in a verified local package. That can make a compatible model runnable on a device where its full files would not fit, with performance determined by the device, storage, context, and model itself.
How Overfit works
1 · Prepare
On Mac, Noema converts a compatible mixture-of-experts GGUF into a `.noema-paged` package. The conversion changes how the files are stored and loaded; it does not retrain, prune, or replace the model.
2 · Set the working set
At launch, Noema decides how much of the model can stay in memory. More available memory means fewer reads from storage; less memory leaves more room for the conversation and the rest of the app.
3 · Load what is needed
When a response needs a part that is not already in memory, Noema reads it from local storage, verifies it, and makes it available to the model.
4 · Run locally
Noema runs the model with its native Apple-silicon runtime and reuses model parts that are already available when possible.
Before you rely on it
The Canary Test runs the exact package on the actual device and reports whether the experience is interactive, slow, better on another device, or unsupported. Results can change when the package, device, storage, or app build changes.
What it measures
Possible result
Advanced behavior
After a successful conversation, Noema can save a local checkpoint and reuse an unchanged system prompt and conversation prefix on a matching future launch.
Checkpoints are tied to the exact model and settings. If a checkpoint cannot be used, Noema falls back to normal prompt processing.
What stays local
If the device is not a good fit, Noema may suggest using a Mac through Constellation. That remains a separate choice.
Safety checks
Noema verifies package structure and file checksums before using model parts. Memory, storage, and thermal safeguards limit work under pressure.
What to expect
A model that fits completely in memory will usually start faster, generate faster, and use less energy. Overfit is useful when access to a larger model matters more than minimum latency.
Supported models
The current runtime supports paged packages for Qwen 3, Qwen 3.5, Qwen 3.6, Gemma 4,Poolside Laguna S 2.1 and Laguna XS 2.1, and DeepSeek V4 Flash 0731. It is not a generic mechanism for dense models or arbitrary GGUF architectures.
DeepSeek V4 Flash is available as a complete package in the Noema collection. For Laguna, download a compatible official GGUF and use Create Paged Package in Stored on macOS. Package creation, context limits, and auxiliary or multimodal workflows can still vary by model.
Pre-built packages are published in the NoemaAI-labs/Noema-Overfit repository on Hugging Face. Poolside Laguna source GGUFs are linked above for Mac package creation.
Noema Overfit
Local execution with measurable limits.
