← Back to community benchmarks
DeepSeek-V4-Flash
Performance
16k
tokens
323.9
PP tok/s
24.5
TG tok/s
50588
TTFT (ms)
146.0
Peak mem (GB)
Hardware
Chip
M3 Ultra (80c)
Memory
512 GB
GPU Cores
80
Software
oMLX
v0.3.9.dev1
macOS
macOS 26.4
Context
16,384
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 162.8 | 13.0 | 145.5 GB | view |
| 4k | 410.7 | 26.7 | 145.8 GB | view |
| 8k | 440.6 | 24.7 | 145.7 GB | view |
| 16k | 323.9 | 24.5 | 146.0 GB | current |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 13.0 | 1.00× |
| 2× | 41.2 | 3.17× |
| 4× | 50.8 | 3.91× |
| 8× | 83.5 | 6.42× |