POCKET-26B
A 26B model that out-runs the top 1-bit 27B — on a CPU, no GPU.
Gemma4-based sparse Mixture-of-Experts · ~4B active per token · same stock llama.cpp, side by side
Gemma4-26B-A4B based 25.2B total ~4B active/token 0 GPU Q4_K_M · 17 GB
starting…
🌳 Bonsai-27B 1-bit dense · runs first
📦 POCKET-26B Gemma4 sparse MoE · runs second
Type a prompt — Bonsai answers first, then POCKET-26B on the same CPU.

What am I looking at?

The exact same prompt is sent to Bonsai-27B (the most-downloaded on-device model, a 1-bit dense build) and to POCKET-26B (our Gemma4-based sparse Mixture-of-Experts, Q4_K_M) — on the same CPU-only box, the same stock llama.cpp. Bonsai runs first, then POCKET. Watch the tok/s on each side.

Why POCKET wins on generation: a dense 27B reads all of its weights every token; POCKET's MoE wakes only ~4B parameters. Fewer bytes moved per token → faster on a CPU. The weaker the hardware, the wider the gap.

Honest notes: sequential race, each model gets the full CPU. tok/s is llama.cpp's measured generation speed. On a shared box a dedicated machine is faster. Prompt processing (not shown) is one place Bonsai can lead on very long inputs. Bonsai is a trademark of its owners; shown here for factual comparison only.