← Back to community benchmarks
Qwen3-Next-80B-A3B-Instruct
Performance
64k
tokens
628.0
PP tok/s
26.1
TG tok/s
104356
TTFT (ms)
46.5
Peak mem (GB)
Hardware
Chip
M2 Ultra (60c)
Memory
64 GB
GPU Cores
60
Software
oMLX
v0.2.6
macOS
macOS 26.3
Context
65,536
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 729.9 | 61.8 | 43.0 GB | view |
| 4k | 848.4 | 59.5 | 43.2 GB | view |
| 8k | 852.2 | 55.4 | 43.4 GB | view |
| 16k | 822.4 | 49.4 | 43.9 GB | view |
| 32k | 752.3 | 38.7 | 44.7 GB | view |
| 64k | 628.0 | 26.1 | 46.5 GB | current |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 61.8 | 1.00× |
| 2× | 98.6 | 1.60× |
| 4× | 146.6 | 2.37× |
| 8× | 203.0 | 3.28× |