← Back to community benchmarks
gpt-oss-120b
Performance
32k
tokens
1,096
PP tok/s
53.6
TG tok/s
29889
TTFT (ms)
117.8
Peak mem (GB)
Hardware
Chip
M3 Ultra (80c)
Memory
512 GB
GPU Cores
80
Software
oMLX
v0.5.1
macOS
macOS 26.3
Context
32,768
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 923.8 | 66.2 | 116.1 GB | view |
| 4k | 1,314 | 64.3 | 116.3 GB | view |
| 8k | 1,296 | 62.1 | 116.6 GB | view |
| 16k | 1,225 | 59.1 | 117.0 GB | view |
| 32k | 1,096 | 53.6 | 117.8 GB | current |
| 64k | 893.7 | 43.3 | 119.5 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 66.2 | 1.00× |
| 2× | 96.7 | 1.46× |
| 4× | 133.2 | 2.01× |
| 8× | 158.4 | 2.39× |