← Back to community benchmarks
gemma-4-31b-it-mxfp8
Performance
8k
tokens
246.7
PP tok/s
17.5
TG tok/s
33206
TTFT (ms)
34.1
Peak mem (GB)
Hardware
Chip
M3 Ultra (60c)
Memory
256 GB
GPU Cores
60
Software
oMLX
v0.3.9
macOS
macOS 26.5
Context
8,192
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 245.2 | 18.4 | 30.9 GB | view |
| 4k | 247.0 | 16.7 | 33.6 GB | view |
| 8k | 246.7 | 17.5 | 34.1 GB | current |
| 16k | 237.7 | 16.6 | 35.3 GB | view |
| 32k | 220.6 | 15.1 | 38.3 GB | view |
| 64k | 192.3 | 12.7 | 44.2 GB | view |
| 128k | 149.9 | 9.5 | 57.0 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 18.4 | 1.00× |
| 2× | 24.0 | 1.30× |
| 4× | 26.5 | 1.44× |