← Back to community benchmarks
Llama-3.2-3B-Instruct
Performance
64k
tokens
227.2
PP tok/s
9.8
TG tok/s
288475
TTFT (ms)
13.8
Peak mem (GB)
Hardware
Chip
M2 Pro (19c)
Memory
32 GB
GPU Cores
19
Software
oMLX
v0.2.10
macOS
macOS 15.7.1
Context
65,536
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 721.6 | 26.1 | 6.5 GB | view |
| 4k | 712.1 | 23.8 | 6.8 GB | view |
| 8k | 629.5 | 21.4 | 7.3 GB | view |
| 16k | 509.0 | 18.1 | 8.2 GB | view |
| 32k | 364.1 | 14.0 | 10.1 GB | view |
| 64k | 227.2 | 9.8 | 13.8 GB | current |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 26.1 | 1.00× |
| 2× | 25.5 | 0.98× |
| 4× | 47.1 | 1.80× |
| 8× | 87.9 | 3.37× |