Maple-Preview

Maple-Preview is an open-source 20B-A1B ternary-weight reasoning large language model optimized for efficient on-device inference. It features a 24-layer, 256-expert architecture designed to deliver state-of-the-art reasoning capabilities while maintaining high performance on consumer hardware like the Mac mini M4.

Maple-Preview is an open-source 20B-A1B ternary-weight reasoning large language model optimized for efficient on-device inference. It features a 24-layer, 256-expert architecture designed to deliver state-of-the-art reasoning capabilities while maintaining high performance on consumer hardware like the Mac mini M4.

• 0120B-A1B ternary-weight architecture
• 02256-expert configuration with 8 active experts
• 03131,072 token context window
• 04High-speed inference (200+ tokens/sec on M4 Mac mini)
• 05Optimized for on-device deployment