← Back to community benchmarks
gemma-4-31b-it-oQ8
Performance
64k
tokens
256.3
PP tok/s
12.2
TG tok/s
255682
TTFT (ms)
45.1
Peak mem (GB)
Hardware
Chip
M3 Ultra (80c)
Memory
512 GB
GPU Cores
80
Software
oMLX
v0.3.8
macOS
macOS 26.4.1
Context
65,536
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 322.1 | 18.2 | 31.8 GB | view |
| 4k | 332.5 | 17.8 | 34.5 GB | view |
| 8k | 331.1 | 17.3 | 35.0 GB | view |
| 16k | 320.9 | 16.5 | 36.1 GB | view |
| 32k | 298.7 | 15.0 | 39.0 GB | view |
| 64k | 256.3 | 12.2 | 45.1 GB | current |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 18.2 | 1.00× |
| 2× | 30.1 | 1.65× |
| 4× | 43.8 | 2.41× |
| 8× | 55.0 | 3.02× |