Benchmarks · community-measured

bonsai-27b-2bit on Apple M2

Real measured local-inference performance for bonsai-27b-2bit on an Apple M2 (24 GB unified memory), from the rapid-mlx community benchmark corpus. Latest submission 2026-08-15, measured on rapid-mlx 0.12.12.

3.8
tok/s decode
25.0
tok/s prefill
86.0 s
first token
11.1 GB
peak memory

A single community run so far — treat it as one data point, not a distribution. Reproduce it below and submit yours to firm it up.

All measurements for this pair

RAMAccelDecodePrefillTTFTPeak memRunsrapid-mlxLatest
24 GB3.8 tok/s25.0 tok/s86.0 s11.1 GB10.12.122026-08-15

Standardized single-request benchmark: 5 measured rounds after 1 discarded warmup, greedy sampling, prefix cache disabled. Raw corpus: /api/benchmarks.

Reproduce this number

Install rapid-mlx, run the same suite on your Mac, and submit the result — it lands on the community leaderboard and updates this page.

$ curl -fsSL https://rapidmlx.com/install.sh | bash
$ rapid-mlx bench bonsai-27b-2bit --tier speed --submit

More

bonsai-27b-2bit model page · Small & curated models · Interactive leaderboard · Alias catalog