← Back to community benchmarks

Huihui-Qwen3.6-35B-A3B-Claude-4.7-Opus-abliterated-v

M3 Max (40c) · 48 GB · 4bit · 2026-04-30
Performance
1k
tokens
1,090
PP tok/s
97.1
TG tok/s
939
TTFT (ms)
18.3
Peak mem (GB)
Hardware
Chip M3 Max (40c)
Memory 48 GB
GPU Cores 40
Software
oMLX v0.3.6
macOS macOS 26.4.1
Context 1,024
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,090 97.1 18.3 GB current
4k 1,443 93.4 19.0 GB view
Batching Results
Batch Size TG tok/s Speedup
97.1 1.00×
154.6 1.59×
228.8 2.36×