← Back to community benchmarks
nightmedia--gpt-oss-120b-heretic-v2-mxfp4-q8-hi
Performance
1k
tokens
977.7
PP tok/s
72.8
TG tok/s
1047
TTFT (ms)
59.5
Peak mem (GB)
Hardware
Chip
M3 Ultra (60c)
Memory
96 GB
GPU Cores
60
Software
oMLX
v0.2.19
macOS
macOS 26.3.1
Context
1,024
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 977.7 | 72.8 | 59.5 GB | current |
| 4k | 1,069 | 71.9 | 59.6 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 72.8 | 1.00× |
| 2× | 94.3 | 1.30× |
| 4× | 137.5 | 1.89× |
| 8× | 175.6 | 2.41× |