AMD Instinct MI250X 128GB vs NVIDIA RTX 5090 for local LLMs
0 reports on the AMD Instinct MI250X 128GB and 170 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 1.8x higher on the AMD Instinct MI250X 128GB (3,277 GB/s on the AMD Instinct MI250X 128GB, 1,792 on the NVIDIA RTX 5090). The estimates below compare them on the models people run most.
- Memory
- 128GB
- Memory bandwidth
- 3,277 GB/s
- FP16 compute
- 383 TFLOPS
- Reports
- 0
- Memory
- 32GB
- Memory bandwidth
- 1,792 GB/s
- FP16 compute
- 419 TFLOPS
- Reports
- 170
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 | AMD Instinct MI250X 128GB | NVIDIA RTX 5090 |
|---|---|---|
| Qwen3.8 27B | 33.4 t/sFP16 | 44.9 t/sQ6_K |
| Qwen3.8 125B · 6B active | 119.8 t/sQ6_K | 20.3 t/sQ2_K, experts in RAM |
| Qwen3.6 35B · 3B active | 120.2 t/sFP16 | 163.0 t/sQ5_K_M |
| DeepSeek V4 Flash 284B · 13B active | 318.2 t/sQ2_K | doesn't fit at these settings |
| Qwen3.6 27B | 33.4 t/sFP16 | 44.9 t/sQ6_K |
| Gemma 4 26B · 4B active | 90.1 t/sFP16 | 114.5 t/sQ6_K |
Change the context, RAM or model in the calculator: AMD Instinct MI250X 128GB · NVIDIA RTX 5090
Every report on each card, with its source: AMD Instinct MI250X 128GB · NVIDIA RTX 5090