LiquidAI LFM2.5-1.2B-Instruct is a 1.2B parameter model from LiquidAI, available at no cost. At this scale, benchmark performance is very limited across reasoning, coding, and instruction following, placing it at the lower end of the benchmarked field. The 32K context window and text-only modality further limit its scope.

For businesses, it is only practical for the simplest, highest-volume text tasks where near-zero cost is the overriding requirement: basic classification, keyword extraction, or simple template filling. It is not suited for any task requiring reliable reasoning, coding, or nuanced language output.

Being free makes it worth evaluating for bulk preprocessing pipelines where a larger model handles exceptions, but teams should expect to route most meaningful tasks to a stronger model. It is not a standalone business tool.

Assessed June 17, 2026

Rankings consider pricing, capabilities, benchmarks, and real-world applicability and are refreshed as new models launch. Feedback?