# Homelabber’s second-gen Optane build uses two RTX 3090s to run 2.8-trillion-parameter Kimi AI model at 3.11 tokens per second

Source: [Latest from Tom's Hardware](https://www.tomshardware.com/tech-industry/artificial-intelligence/homelabbers-second-gen-optane-build-runs-2-8-trillion-parameter-kimi-k3-at-3-11-tokens-per-second-after-adding-two-rtx-3090s-moving-nearly-half-the-model-into-a-ram-drive-boosted-performance-by-41-percent) · 2026-11-21 · Industry
Page: https://themorning.dev/story/homelabber-s-second-gen-optane-build-uses-two-rtx-3090s-to-run-2-8-trillion-para-a8b9571f5922480c
Kind: other · Impact: low

## In brief

- A homelabber upgraded a second-generation Optane-based AI build by adding two RTX 3090 GPUs.
- The setup runs the 2.8-trillion-parameter Moonshot Kimi K3 model at 3.11 tokens per second.
- Moving almost half of the model onto a RAM drive reportedly boosted performance by 41 percent.

## Why it matters

It shows how a consumer-grade homelab with multiple GPUs and optimized memory placement can run a very large model, though the reported speed is modest and the details come from a single hobbyist build.

## Versions

- Kimi K3

## Timeline

- 2026-11-21: Tom's Hardware reports on the homelabber's follow-up post detailing the upgraded build

## Coverage

- [Homelabber’s second-gen Optane build uses two RTX 3090s to run 2.8-trillion-parameter Kimi AI model at 3.11 tokens per second](https://www.tomshardware.com/tech-industry/artificial-intelligence/homelabbers-second-gen-optane-build-runs-2-8-trillion-parameter-kimi-k3-at-3-11-tokens-per-second-after-adding-two-rtx-3090s-moving-nearly-half-the-model-into-a-ram-drive-boosted-performance-by-41-percent) — Latest from Tom's Hardware, 2026-11-21

Tags: homelab, local-llm, optane, gpu, kimi

---

Brief generated by AI (claude-haiku-5-5) from the reporting listed under Coverage. Headlines belong to their publishers. The Morning Dev – https://themorning.dev
