← Back to community benchmarks
gpt-oss-20b-MXFP4-Q8
Performance
16k
tokens
1,156
PP tok/s
74.9
TG tok/s
14171
TTFT (ms)
11.9
Peak mem (GB)
Hardware
Chip
M3 Ultra (60c)
Memory
256 GB
GPU Cores
60
Software
oMLX
v0.2.7
macOS
macOS 26.3
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 1,076 | 111.8 | 11.4 GB | view |
| 4k | 1,247 | 103.2 | 11.5 GB | view |
| 8k | 1,239 | 91.5 | 11.7 GB | view |
| 16k | 1,156 | 74.9 | 11.9 GB | current |
| 32k | 1,014 | 55.5 | 12.3 GB | view |
| 64k | 806.0 | 36.9 | 13.1 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 111.8 | 1.00× |
| 2× | 154.0 | 1.38× |
| 4× | 201.2 | 1.80× |
| 8× | 251.5 | 2.25× |