Compare
Compare local AI hardware with workload-aware output.
RTX 4090 24GB wins for coding in balanced mode
Based on model fit, speed, and quality across top recommendations.
MacBook Pro M3 Pro 18GB
SQwen 3.5 9B
llama.cppq4-k-mRuns well
10.5 GB / 13.0 GB
21.4 tok/s34K ctx
ACodeGeeX 4 9B
llama.cppq4-k-mRuns well
8.9 GB / 13.0 GB
21.8 tok/s121K ctx
AGemma 4 E4B
llama.cppq4-k-mRuns well
9.0 GB / 13.0 GB
18.3 tok/s65K ctx
Quick comparison
| Metric | MacBook Pro M3 Pro 18GB | RTX 4090 24GB |
|---|---|---|
| Models that fit | 3 | 3 |
| Avg decode tok/s | 20.5 | 37.9 |
| Best grade score | 94 | 93 |