← Back to community benchmarks
GLM-4.7
Performance
8k
tokens
222.8
PP tok/s
13.0
TG tok/s
36767
TTFT (ms)
271.1
Peak mem (GB)
Hardware
Chip
M3 Ultra (80c)
Memory
512 GB
GPU Cores
80
Software
oMLX
v0.2.6
macOS
macOS 26.3
Context
8,192
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 235.1 | 16.6 | 268.3 GB | view |
| 4k | 241.6 | 14.8 | 269.7 GB | view |
| 8k | 222.8 | 13.0 | 271.1 GB | current |
| 16k | 189.6 | 10.4 | 274.0 GB | view |
| 32k | 145.2 | 7.7 | 279.8 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 16.6 | 1.00× |
| 2× | 23.4 | 1.41× |
| 4× | 28.0 | 1.69× |
| 8× | 34.6 | 2.08× |