Motif 3
Motif Technologies · 314.8B parameters
Motif Technologies' released sparse mixture-of-experts language model, distinct from Motif-3-Beta. The MIT-licensed release supports native 262,144-token text context and agentic tool use; its BF16 checkpoint contains 314,841,775,750 parameters. The model card reports approximately 13.2B active parameters per token, recorded at that published precision.