Unified MemoryBenchmarks / Which Small AI Model Should You Run on a Mac? We Ran 312 Tests / Models
Granite 4.2 8B
ibm-granite/granite-4.2-8b-q4-mlx · MLX · mlx_lm.server 0.32.2
#3 of 8
overall, text jobs
84.4
quality score
42.6
median tokens/s
8.13 GB
RAM with 8K context filled
4.61 GB
download size
Score per job
| Job | Accuracy | Pass | Easy | Medium | Hard | Median latency | Tok/s | Job time | Errors |
|---|---|---|---|---|---|---|---|---|---|
| Assistant actions | 76.7 | 64.0% | 86.7% | 64.7% | 44.4% | 1.41 s | 33.0 | 1m 28s | 0 |
| Ask my documents | 86.3 | 82.5% | 100.0% | 84.6% | 66.7% | 4.52 s | 39.3 | 3m 37s | 0 |
| Rewrite and proofread | 82.9 | 47.6% | 71.4% | 50.0% | 21.4% | 2.74 s | 42.6 | 2m 26s | 0 |
| Summarize + extract | 89.5 | 45.5% | 60.0% | 38.5% | 37.5% | 3.70 s | 43.8 | 3m 24s | 0 |
| Translate | 79.1 | 55.0% | 69.2% | 53.8% | 42.9% | 2.08 s | 41.9 | 3m 33s | 0 |
| Sort and triage | 92.0 | 91.1% | 94.1% | 94.7% | 85.0% | 0.75 s | 48.1 | 40.7 s | 0 |
Easy / medium / hard columns are pass rates within each difficulty tier.
Memory and loading
| Download size | 4.61 GB |
| Loaded, idle | 4.85 GB |
| In use (8K context) | 5.54 GB |
| 8K context filled | 8.13 GB |
| Server peak during the jobs (incl. reserved context) | 10.01 GB |
| Load time | 3m 32s |
Measured on Apple M4 Max, 36 GB, macOS 27.0.1. How memory was measured.