← Back to community benchmarks
Qwen3-Coder-Next
Performance
32k
tokens
1,598
PP tok/s
37.2
TG tok/s
20501
TTFT (ms)
81.8
Peak mem (GB)
Hardware
Chip
M3 Ultra (80c)
Memory
256 GB
GPU Cores
80
Software
oMLX
v0.2.10
macOS
macOS 26.3
Context
32,768
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 1,192 | 62.3 | 80.1 GB | view |
| 4k | 1,825 | 62.3 | 80.3 GB | view |
| 8k | 1,924 | 58.1 | 80.5 GB | view |
| 16k | 1,827 | 54.3 | 80.9 GB | view |
| 32k | 1,598 | 37.2 | 81.8 GB | current |
| 64k | 1,241 | 24.4 | 83.6 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 62.3 | 1.00× |
| 2× | 109.2 | 1.75× |
| 4× | 174.4 | 2.80× |
| 8× | 269.0 | 4.32× |