Compare
Compare local AI hardware with workload-aware output.
NVIDIA B200 180GB wins for coding in balanced mode
Based on model fit, speed, and quality across top recommendations.
NVIDIA B200 180GB
WinnerSDevstral 2 123B Instruct
llama.cppq6-kRuns well
125.1 GB / 180.0 GB
76.1 tok/s179K ctx
SQwen 3.5 122B A10B
llama.cppq6-kRuns well
121.4 GB / 180.0 GB
211.0 tok/s131K ctx
SMistral Small 4 119B
llama.cppq6-kRuns well
121.9 GB / 180.0 GB
228.8 tok/s189K ctx
Quick comparison
| Metric | NVIDIA B200 180GB | AMD Instinct MI300X 192GB |
|---|---|---|
| Models that fit | 3 | 3 |
| Avg decode tok/s | 172.0 | 105.8 |
| Best grade score | 99 | 98 |