← Back to community benchmarks
Jundot-Qwen3.6-27B-oQ8-mtp
Performance
16k
tokens
365.5
PP tok/s
28.7
TG tok/s
44826
TTFT (ms)
32.8
Peak mem (GB)
Hardware
Chip
M3 Ultra (80c)
Memory
512 GB
GPU Cores
80
Software
oMLX
v0.3.9.dev2
macOS
macOS 26.3.1
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 354.4 | 30.3 | 28.8 GB | view |
| 4k | 396.0 | 30.8 | 30.3 GB | view |
| 8k | 389.1 | 30.2 | 31.3 GB | view |
| 16k | 365.5 | 28.7 | 32.8 GB | current |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 30.3 | 1.00× |
| 2× | 38.5 | 1.27× |
| 4× | 52.2 | 1.72× |