ZMIME

Aurora2-400M-A20M vs LFM2 24B A2B

Compare documented capabilities and access. A dash means a comparable figure is unavailable.

Model specifications
SpecificationAurora2-400M-A20MAuroraAI ResearchLFM2 24B A2BLiquid AI
TaskBase language modeling—
Published capabilitiesEnglish base model pretrained on 100 billion tokens with one shared expert and 32 routed experts per layer.Efficient open Liquid MoE instruct model for on-device chat and tool calling
Model size400M total / approximately 19M active parameters—
AccessPublic Transformers checkpoint for completion and adaptation—
SourcesModel weights—
InputsTextText
Context window—32,768 tokens
Input price per million tokens—$0.03
Output price per million tokens—$0.12
WeightsOpen weights · Apache-2.0Open weights · LFM Open License v1.0
ReleasedNov 18, 2026Feb 24, 2026