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 24GB
SQwen 3.5 9B
llama.cppq6-kRuns well
13.1 GB / 17.3 GB
10.4 tok/s47K ctx
ANemotron Nano 9B v2
llama.cppq4-k-mRuns well
11.4 GB / 17.3 GB
13.3 tok/s54K ctx
ACodeGeeX 4 9B
llama.cppq6-kRuns well
11.5 GB / 17.3 GB
10.6 tok/s131K ctx
Quick comparison
| Metric | MacBook Pro M3 24GB | RTX 4090 24GB |
|---|---|---|
| Models that fit | 3 | 3 |
| Avg decode tok/s | 11.4 | 37.9 |
| Best grade score | 92 | 93 |