The job, not the leaderboard.

Buying advice organized by what you're actually trying to do.

• Under $1k

  2026-05-01

  Local LLM rig under $1k (2026)

  A $1k rig won't run a 70B model at production speed. It will run 14B-class models in Q8 with full context, or 32B in Q4 — fast enough for daily dev work, no cloud bill.

  Run 14-32B models at home on a budget rig — small models, full context.

• Under $10k

  2026-05-01

  Serious local LLM workstation under $10k (2026)

  Two RTX 5090s, a Threadripper for PCIe lanes, and 128 GB of DDR5. 64 GB of aggregate VRAM at the lowest cost-per-GB on this tier — and it sits on your desk.

  Run 70B+ models comfortably, multi-GPU agentic workflows, or LoRA fine-tunes — all locally.

• Under $4k

  2026-05-01

  Image and video generation rig under $4k (2026)

  Single RTX 5090, 64 GB DDR5, 4 TB NVMe. Built for sustained image and short-video runs — Flux, SDXL, Wan 2.2 — without the cloud bill or the queue.

  Generate images and short videos locally — Flux, SDXL, Wan 2.2 — at production iteration speeds.

• Datacenter

  2026-05-01

  Single-rack production inference blueprint (2026)

  Spec a single 19-inch rack for production LLM inference. HGX B200 in OEM 4U as the default, DGX B200 if you want the appliance, GB200 NVL72 if your facility is liquid-ready.

  Spec a single rack for company-tier LLM inference. Decide air vs liquid; own vs rent.

• Under $4k

  2026-04-24

  Local LLM rig under $4k (2026)

  The minimum-viable workstation for serious local inference: single 5090, 64 GB system RAM, fast NVMe, and a case that holds up under sustained load.

  Run 70B-class models at home, without the cloud bill.