Skip to main content

Blog & resources

Notes from the lab.

Engineering notes, launch news, and plain-English explainers on durable long-context AI — method, measurements, and the occasional argument for converting models instead of retraining them.

Featured · announcements

Your model, converted: sub-quadratic inference for the models you already trust

lab358 converts open-weight LLMs into a natively sparse top-k attention architecture served by ANN retrieval — same benchmark quality, sub-quadratic long-context cost. Coming to AWS Marketplace.

Michael Beck· June 4, 2026
More on the way.

Cross-family conversion benchmarks — and scale-up beyond the 3B target — are in progress. We'll publish them here: measured, not projected.

Request access