← Back to community benchmarks
gemma-4-26B-A4B-it
Performance
32k
tokens
1,440
PP tok/s
49.7
TG tok/s
22750
TTFT (ms)
50.0
Peak mem (GB)
Hardware
Chip
M3 Ultra (60c)
Memory
256 GB
GPU Cores
60
Software
oMLX
v0.3.8
macOS
macOS 26.3
Context
32,768
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 1,214 | 63.8 | 47.8 GB | view |
| 4k | 1,545 | 62.2 | 48.5 GB | view |
| 8k | 1,608 | 60.1 | 48.6 GB | view |
| 16k | 1,569 | 56.2 | 49.1 GB | view |
| 32k | 1,440 | 49.7 | 50.0 GB | current |
| 64k | 1,195 | 39.8 | 52.0 GB | view |
| 128k | 865.4 | 23.2 | 56.2 GB | view |
| 195k | 655.3 | 18.0 | 61.2 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 63.8 | 1.00× |
| 2× | 58.7 | 0.92× |
| 4× | 95.4 | 1.50× |
| 8× | 77.3 | 1.21× |