← Back to community benchmarks
gemma-4-31B-it-The-DECKARD-HERETIC-UNCENSORED-Thinking-qx86-hi
Performance
16k
tokens
203.7
PP tok/s
7.2
TG tok/s
80430
TTFT (ms)
35.2
Peak mem (GB)
Hardware
Chip
M3 Ultra (60c)
Memory
256 GB
GPU Cores
60
Software
oMLX
v0.3.7
macOS
macOS 26.2
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 246.6 | 20.0 | 28.4 GB | view |
| 4k | 250.7 | 19.1 | 30.2 GB | view |
| 8k | 248.1 | 5.4 | 31.5 GB | view |
| 16k | 203.7 | 7.2 | 35.2 GB | current |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 20.0 | 1.00× |
| 2× | 19.3 | 0.97× |
| 4× | 26.3 | 1.31× |
| 8× | 93.7 | 4.69× |