ZeptoLM35 M · Q3_K_M · 22 MB

Booting…
Variant A + LoRA SFT, runs entirely in your browser via wllama (llama.cpp WASM).
Preparing…
Metrics & info

Status

Initializing…

Live

Disk22 MB
tok/s decode—
TTFT—
Total tokens0

vs LFM2-350M (Mac CPU, Q4_K_M)

disk8.4× smaller
RAM4× less
tok/s5× faster
tool_match1.00 vs 0.47
Loading 22 MB GGUF weights from this page's directory…