Welcome! Type "help" for available commands.
$
Summary
Maple-Preview is an open-source 20B-A1B ternary-weight reasoning LLM by DeepGrove, with a 5.31 GB checkpoint and 131K token context.
It scores 87.5 on AIME 26 and 73.5 on GPQA-D, running at 200+ tokens/s on a Mac mini M4 and 127 tokens/s on an iPhone. Weights are on Hugging Face, with local run instructions on GitHub and a chat demo at deepgrove.ai.