RTX 5090 vs AMD Strix Halo 128GB
For running local LLMs · 100 reports across 15 models
Highest reported generation speed by model family
These maxima come from different community setups. Model size, quantization, GPU count, offloading and concurrency may differ. They do not establish which GPU is faster for the same workload. Check the comparison conditions →
| Model | RTX 5090 | AMD Strix Halo 128GB |
|---|---|---|
| Qwen3.8up to 2400B | 220.0n=35 | 153.3n=18 |
| DeepSeek V4 Flashup to 285B | 120.0n=10 | 35.0n=4 |
| Qwen3.6up to 35B | 215.1n=6 | 50.0n=8 |
| Gemma 4up to 31B | 578.0n=4 | — |
| Museup to 30B | 253.0n=4 | — |
| Qwen3.5up to 122B | —n=1 | 49.0n=1 |
| GLM-4.7 | — | 40.0n=1 |
| GLM-5.2up to 744B | —n=1 | — |
| GLM-5.3up to 321B | 24.5n=1 | — |
| Kimi K2.5 | —n=1 | — |
| LFM2.5up to 3B | — | 113.0n=1 |
| Nex-N2.5-mini | — | 76.9n=1 |
| Qwen2.5up to 32B | —n=1 | — |
| Qwen3up to 235B | 159.0n=1 | — |
| Qwen3-Coder-Next | — | 36.8n=1 |