Lari Live · Approved snapshots
Lari Live: a public research observatory.
Lari Live is the evidence trail for a different AI model architecture in development. It presents reviewed public records as static snapshots; it has no connection to LARI’s private runtime and updates only when a reviewed snapshot is built and published. See what has been tested, what remains a candidate or negative result, and what the evidence does not establish.
Approved evidence is available.
This snapshot was generated Oct 31, 2026 from approved records.
Architecture context
Evidence belongs to a component.
Approved records will identify which architectural component they test, then connect the result to its evidence and limitations. Approved records link tested components to their evidence and limitations.
Explore LARI’s architectureApproved collections
Methods, outcomes,
and boundaries.
Published experiments include their evaluation method and limitations. Capability claims and development milestones appear only after record-level review. Unsuccessful and inconclusive results belong in the record when they have been approved.
How evidence is labeledSnapshot history
Freshness is explicit.
The current approved snapshot was generated Oct 31, 2026. Historical comparisons will appear only when prior approved snapshots are preserved; there are no synthetic history entries.
The data feed is static JSON. A new research event does not appear here until its record passes review, export validation, and the website build.
Comparable measurements
Metrics stay in context.
Only explicitly approved measurements with their conditions are shown. Different tasks, models, hardware, or datasets are not combined into a progress score. No chart is rendered when comparable approved measurements are unavailable.
No approved public records
No comparable metrics are approved yet.
The observatory will show individual measures with their dates, denominators, and conditions after review; it will not synthesize a score from incompatible benchmarks.
Snapshot history
Discuss this page
Read and join this page’s discussion here. Sign in to GitHub when you want to reply or react.
Loading discussion from GitHub…
The embedded discussion could not be loaded. Browse discussions on GitHub ↗