← Back to community benchmarks
gemma-4-31B-it-oQ8
Performance
16k
tokens
322.8
PP tok/s
16.7
TG tok/s
50751
TTFT (ms)
34.7
Peak mem (GB)
Hardware
Chip
M3 Ultra (80c)
Memory
512 GB
GPU Cores
80
Software
oMLX
v0.3.10
macOS
macOS 26.5
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 319.4 | 18.3 | 31.8 GB | view |
| 4k | 337.5 | 17.9 | 33.0 GB | view |
| 8k | 333.4 | 17.5 | 33.5 GB | view |
| 16k | 322.8 | 16.7 | 34.7 GB | current |
| 32k | 300.2 | 15.4 | 37.0 GB | view |
| 64k | 257.4 | 13.1 | 41.8 GB | view |
| 128k | 189.8 | 10.0 | 51.3 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 18.3 | 1.00× |
| 2× | 30.3 | 1.66× |
| 4× | 43.9 | 2.40× |
| 8× | 53.5 | 2.92× |