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.
RX 7900 XT 20GB
SQwen 3.5 9B
llama.cppq6-kRuns well
12.5 GB / 20.0 GB
73.4 tok/s71K ctx
ACodeGeeX 4 9B
llama.cppq6-kRuns well
10.9 GB / 20.0 GB
74.7 tok/s131K ctx
ANemotron Nano 9B v2
llama.cppq6-kRuns well
12.7 GB / 20.0 GB
73.4 tok/s64K ctx
Quick comparison
| Metric | RX 7900 XT 20GB | RTX 4090 24GB |
|---|---|---|
| Models that fit | 3 | 3 |
| Avg decode tok/s | 73.8 | 37.9 |
| Best grade score | 96 | 93 |