Provider pricing
Liquid AI AI API pricing
Liquid AI builds Liquid Foundation Models (LFM-1B, LFM-3B, LFM-40B MoE) using a novel non-Transformer state-space architecture. LFMs deliver competitive quality with dramatically lower memory footprint and longer effective context — making them well-suited for edge, on-device, and high-throughput serving.
Founded2023
HeadquartersCambridge, MA
Tracked modelsDirectory
Best for
Edge AI, on-device assistants, and teams needing efficient inference on constrained hardware.
Strengths
- Non-Transformer architecture: lower memory, longer effective context
- LFM-3B is competitive with much larger Transformer models on benchmarks
- On-device deployment supported (mobile, embedded)
- MIT-spinout team with deep state-space-model expertise