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.
GTX 1650 4GB
CStarCoder2 3B
llama.cppq3-k-mRuns well
3.1 GB / 4.0 GB
40.5 tok/s56K ctx
AQwen 2.5 Coder 1.5B
llama.cppq6-kRuns well
3.0 GB / 4.0 GB
21.0 tok/s33K ctx
CStarCoder2 3B
llama.cppq4-k-mTight fit
3.6 GB / 4.0 GB
38.0 tok/s16K ctx
Quick comparison
| Metric | GTX 1650 4GB | RTX 4090 24GB |
|---|---|---|
| Models that fit | 3 | 3 |
| Avg decode tok/s | 33.2 | 37.9 |
| Best grade score | 71 | 93 |