← Back to community benchmarks

Qwen3.5-27B-Claude-4.6-Opus-Distilled

M3 Ultra (80c) · 512 GB · 4bit · 2026-03-10
Performance
32k
tokens
378.3
PP tok/s
25.5
TG tok/s
86624
TTFT (ms)
19.2
Peak mem (GB)
Hardware
Chip M3 Ultra (80c)
Memory 512 GB
GPU Cores 80
Software
oMLX v0.2.6
macOS macOS 26.2
Context 32,768
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 384.9 36.0 15.8 GB view
4k 415.0 33.8 16.1 GB view
8k 414.3 32.5 16.6 GB view
16k 400.8 29.4 17.5 GB view
32k 378.3 25.5 19.2 GB current
Batching Results
Batch Size TG tok/s Speedup
36.0 1.00×
64.7 1.80×
88.1 2.45×