← Back to community benchmarks
Laguna-S-2.1-oQ5e
Performance
16k
tokens
787.0
PP tok/s
51.2
TG tok/s
20819
TTFT (ms)
74.8
Peak mem (GB)
Hardware
Chip
M3 Ultra (80c)
Memory
512 GB
GPU Cores
80
Software
oMLX
v0.5.3
macOS
macOS 26.5.2
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 848.7 | 55.7 | 73.6 GB | view |
| 4k | 855.4 | 54.7 | 73.9 GB | view |
| 8k | 825.1 | 53.3 | 74.2 GB | view |
| 16k | 787.0 | 51.2 | 74.8 GB | current |
| 32k | 716.5 | 46.5 | 75.9 GB | view |
| 64k | 605.5 | 40.5 | 78.3 GB | view |
| 128k | 458.8 | 32.7 | 83.0 GB | view |
| 195k | 363.6 | 26.6 | 89.4 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 55.7 | 1.00× |
| 2× | 87.3 | 1.57× |
| 4× | 125.2 | 2.25× |
| 8× | 172.5 | 3.10× |