Maple
Sub-2-BPW Moe Pruning to Run Gemma 4 26B at 45 TPS on a 6GB GPU
Maple launched on 8 November 2026. It ranked #24 on Daily Launch on 9 November 2026 with a Launch Score of 61, from 20 points on Show HN. It was new in the ranking that day.
What is Maple?
The MAPLE mathematical formulation operates on the fundamental routing mechanism shared by all Sparse Mixture-of-Experts (SMoE) transformers. It is architecturally portable to any MoE model family, including: - Liquid AI LFM-2 MoE / LFM 24B (fine-grained hybrid conv/attention MoE) - Mistral / Mixtral 8x7B & 8x22B (standard top-2 routed MoE) - DeepSeek-V2 / DeepSeek-V3 / DeepSeek MoE (fine-grained routed experts with shared expert isolation) - Qwen 2.5 MoE (e.g.
From its README on GitHub
Maple is a launch on Show HN. It is written in Python.
Launch timeline
Launch Score, last 7 days
Ranked on one day so far: 61 of 100 on 9 November.
#58 of 228 in the roundup for the week of 9-15 November 2026.