← Back to community benchmarks

Qwen3.5-35B-A3B

M1 Ultra (48c) · 128 GB · 4bit · 2026-04-17
Performance
16k
tokens
835.1
PP tok/s
54.2
TG tok/s
19620
TTFT (ms)
21.0
Peak mem (GB)
Hardware
Chip M1 Ultra (48c)
Memory 128 GB
GPU Cores 48
Software
oMLX v0.3.6
macOS macOS 26.4.1
Context 16,384
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 651.6 69.8 19.2 GB view
4k 840.6 64.5 20.0 GB view
8k 858.5 60.3 20.4 GB view
16k 835.1 54.2 21.0 GB current
Batching Results
Batch Size TG tok/s Speedup
69.8 1.00×
110.1 1.58×
150.6 2.16×
189.9 2.72×