Llama 3.1 EVR-1 Maano 8B base, custom EVR, separate pp512 and tg128 tests

on M4 16GB · Evrmind custom Metal binary / llama.cpp CLI

Oct 11, 2026

Throughput

8.2 t/s gen · 9.1 tokens/s prompt processing (prefill)Prompt processing measures input prefill speed; it does not tell you the time to first token.

Quant

EVR-1 custom compression (~3-bit weights) (GGUF)

System RAM

16 GB

Summary

User reports Llama 3.1 8B at 8.2 t/s generation and 9.1 t/s prompt processing on an M4 16GB Mac mini. Setup is Evrmind's custom Metal binary with EVR-1 compression at roughly 3 bits per weight and full GPU offload. The figures come from separate pp512 and tg128 phase tests, not a paired 512-input/128-output run; the source table lists 9.11 and 8.16 t/s.