← Back to community benchmarks
gemma-4-31b
Performance
8k
tokens
179.5
PP tok/s
23.4
TG tok/s
45644
TTFT (ms)
20.7
Peak mem (GB)
Hardware
Chip
M2 Ultra (76c)
Memory
192 GB
GPU Cores
76
Software
oMLX
v0.3.8
macOS
macOS 26.3.1
Context
8,192
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 169.9 | 26.1 | 17.8 GB | view |
| 4k | 179.9 | 24.8 | 20.2 GB | view |
| 8k | 179.5 | 23.4 | 20.7 GB | current |
| 16k | 175.5 | 20.9 | 22.1 GB | view |
| 32k | 167.1 | 16.8 | 24.8 GB | view |
| 64k | 145.2 | 12.4 | 31.4 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 26.1 | 1.00× |
| 2× | 33.7 | 1.29× |
| 4× | 68.4 | 2.62× |
| 8× | 137.1 | 5.25× |