← Back to community benchmarks

MiniMax-M2.5

M3 Ultra (80c) · 512 GB · 6bit · 2026-03-10
Performance
64k
tokens
292.6
PP tok/s
10.9
TG tok/s
224013
TTFT (ms)
189.6
Peak mem (GB)
Hardware
Chip M3 Ultra (80c)
Memory 512 GB
GPU Cores 80
Software
oMLX v0.2.6
macOS macOS 26.3
Context 65,536
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 631.8 39.5 174.0 GB view
4k 726.3 34.4 175.0 GB view
8k 675.2 29.6 175.9 GB view
16k 575.5 23.7 177.8 GB view
32k 437.6 16.3 181.7 GB view
64k 292.6 10.9 189.6 GB current
Batching Results
Batch Size TG tok/s Speedup
39.5 1.00×
58.2 1.47×
74.8 1.89×
105.6 2.67×