← Back to community benchmarks
gemma-4-26B-A4B-it-qat-OptiQ
Performance
32k
tokens
1,352
PP tok/s
62.0
TG tok/s
24237
TTFT (ms)
22.0
Peak mem (GB)
Hardware
Chip
M3 Ultra (60c)
Memory
256 GB
GPU Cores
60
Software
oMLX
v0.5.1
macOS
macOS 26.5.2
Context
32,768
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 1,925 | 85.0 | 20.1 GB | view |
| 4k | 1,725 | 82.0 | 20.4 GB | view |
| 8k | 1,635 | 77.5 | 21.1 GB | view |
| 16k | 1,516 | 72.5 | 21.3 GB | view |
| 32k | 1,352 | 62.0 | 22.0 GB | current |
| 64k | 1,118 | 47.5 | 23.9 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 85.0 | 1.00× |
| 2× | 143.2 | 1.68× |
| 4× | 225.5 | 2.65× |
| 8× | 287.0 | 3.38× |