# A user on X has shared some benchmarks of his test suite, showcasing whether it is possible to run large AI models at home.

*semiconductor · news · 2026-04-19 · Wccftech*

## Key points

- A single NVIDIA RTX Pro 6000 Blackwell matches four RTX 5090s in AI inference speed.
- The RTX Pro 6000 uses 600W peak, just one quarter the power of four RTX 5090s.
- Four RTX 5090 GPUs cost $14,000, while one RTX Pro 6000 costs $9,500.
- Multi-GPU setups encounter overheads reducing effective performance compared to a single GPU Blackwell setup.
- DGX Spark offers much lower power consumption but significantly trails in AI inference speed.

**Companies:** NVIDIA
**Countries:** United States

[Read the full story on Wccftech](https://wccftech.com/nvidia-rtx-pro-6000-matches-four-rtx-5090s-on-230b-ai-model-draws-quarter-power/)

---

Canonical: https://newsio.io/n/3bc34403-98c3-4d3b-9144-3a8460eeff11/a-user-on-x-has-shared-some-benchmarks-of-his-test-suite-showcasing-whether-it-i
Summarized by Newsio from Wccftech. https://newsio.io/how-it-works