A compact, open-weight model from LiquidAI built on the LFM2.5 architecture, handling up to 128k tokens of context. Its relatively small 8B active parameter footprint makes it accessible for local deployment, though specific capability strengths beyond text processing aren't well documented publicly.