NVIDIA H200 vs NVIDIA RTX 5090 for local LLMs
2 reports on the NVIDIA H200 and 161 on the NVIDIA RTX 5090
Which is faster for local LLMs?
Nobody has reported the same model at the same quant on both cards yet, so there's no measured head-to-head. Memory bandwidth, which caps how fast a card can write, is 2.7x higher on the NVIDIA H200 (4,800 GB/s on the NVIDIA H200, 1,792 on the NVIDIA RTX 5090). The estimates below compare them on the models people run most.
NVIDIA H200
- Memory
- 141GB
- Memory bandwidth
- 4,800 GB/s
- FP16 compute
- 989 TFLOPS
- Reports
- 2
NVIDIA RTX 5090
- Memory
- 32GB
- Memory bandwidth
- 1,792 GB/s
- FP16 compute
- 419 TFLOPS
- Reports
- 161
Estimated by the calculator
The quant the calculator recommends for each card and its estimated speed at a 32,768-token context with 32 GB of system RAM, from memory bandwidth and the model's shape.
| Model | NVIDIA H200 | NVIDIA RTX 5090 |
|---|---|---|
| Qwen3.8 27B | 48.9 t/sFP16 | 44.9 t/sQ6_K |
| Qwen3.8 125B · 6B active | 191.1 t/sQ6_K | 20.3 t/sQ2_K, experts in RAM |
| Qwen3.6 35B · 3B active | 176.0 t/sFP16 | 163.0 t/sQ5_K_M |
| DeepSeek V4 Flash 284B · 13B active | 362.5 t/sQ3_K_S | doesn't fit at these settings |
| Qwen3.6 27B | 48.9 t/sFP16 | 44.9 t/sQ6_K |
| Gemma 4 26B · 4B active | 132.0 t/sFP16 | 114.5 t/sQ6_K |
Change the context, RAM or model in the calculator: NVIDIA H200 · NVIDIA RTX 5090
Every report on each card, with its source: NVIDIA H200 · NVIDIA RTX 5090