Mark on Mythryl: “I tested Bonsai 2 27B on a 128GB M3 Max... slow as or slower than…”

I tested Bonsai 2 27B on a 128GB M3 Max... slow as or slower than Qwen 3.8 27B Q8 (prefill is painful on dense models on a Apple silicon) as the performance drops off significantly with increasing context length. Most damning, non-thinking performance was significantly worse.

Sung Kim@sungkim.bsky.social

PrismML's Ternary Bonsai 2 27B Based on Qwen3.8 27B, Bonsai 2 27B is 9x smaller than its full-precision counterpart while retaining 98.2% of its aggregate benchmark performance.

19/9/26, 3:35 pm0 reposts0 likes

Post