Compare
Compare local AI hardware with workload-aware output.
RX 7900 XTX 24GB wins for coding in balanced mode
Based on model fit, speed, and quality across top recommendations.
RX 7900 XTX 24GB
WinnerSDevstral Small 2 24B Instruct
llama.cppQ4_K_MTight fit
20.4 GB / 24.0 GB
50.8 tok/s40K ctx
SCodestral 2 25.08
llama.cppQ4_K_MRuns well
19.2 GB / 24.0 GB
52.0 tok/s48K ctx
SQwen 3.6 27B
llama.cppQ4_K_MTight fit
20.7 GB / 24.0 GB
29.8 tok/s69K ctx
Quick comparison
| Metric | RX 7900 XTX 24GB | RTX 4090 24GB |
|---|---|---|
| Models that fit | 3 | 3 |
| Avg decode tok/s | 44.2 | 34.0 |
| Best grade score | 93 | 93 |