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 36GB
SQwen 3.6 27B
llama.cppq4-k-mTight fit
22.2 GB / 25.9 GB
5.5 tok/s76K ctx
ACodestral 2 25.08
llama.cppq4-k-mRuns well
20.6 GB / 25.9 GB
8.2 tok/s51K ctx
SDevstral Small 2 24B Instruct
llama.cppq4-k-mTight fit
21.9 GB / 25.9 GB
8.0 tok/s43K ctx
Quick comparison
| Metric | MacBook Pro M3 Pro 36GB | RTX 4090 24GB |
|---|---|---|
| Models that fit | 3 | 3 |
| Avg decode tok/s | 7.2 | 37.9 |
| Best grade score | 88 | 93 |