Budget lab
Budget AI Lab
The cheapest all-new-parts build that still runs a useful local model. For testing the waters before committing to a bigger rig.
priceentry costsimplicity
updated 16 Sept 2026Total price
$1,748
VRAM
16 GB
RAM
64 GB
Storage
2 TB
Est. power draw
352 W
AI capability
Roughly ~6B params in fp16, ~13B in 8-bit, or ~24B in 4-bit quantization.
small-model inference
Components
- GPUNVIDIA GeForce RTX 4060 Ti 16GBmemory: 16 GB GDDR6 · bandwidth: 288 GB/s · interface: PCIe 4.0 x8$469Check price ↗
- CPUAMD Ryzen 5 7600cores: 6 · threads: 12 · clock: 5.1 GHz boost$179Check price ↗
- MotherboardASUS ProArt X670E-Creator (AM5)chipset: X670E · memory: 4 x DDR5, up to 192 GB · pcie: 2 x PCIe 5.0 x16 (x8/x8)$300Check price ↗
- RAMG.Skill 64 GB (2x32) DDR5-6000 CL30capacity: 64 GB (2 x 32) · type: DDR5-6000 · ecc: non-ECC$175Check price ↗
- StorageSamsung 990 PRO 2 TB NVMecapacity: 2 TB · bus: PCIe 4.0 NVMe · read: 7450 MB/s$130Check price ↗
- PSUCorsair RM850x (ATX 3.1) 850 W 80+ Goldcapacity: 850 W · rating: 80+ Gold · standard: ATX 3.1 / PCIe 5.1$130Check price ↗
- CoolingArctic Liquid Freezer III 360 (AIO)type: liquid AIO 360 mm · dissipation: up to 350 W · noise: 22-33 dBA$110Check price ↗
- NetworkingIntel I226-V 2.5GbE (onboard-class NIC)speed: 2.5 GbE · ports: 1 · bus: PCIe 3.0 x1
- ChassisFractal Define 7 XL (ATX tower)form factor: E-ATX tower · gpu clearance: 2 triple-slot GPUs realistic · drive bays: up to 18 (with kits)$220Check price ↗
Component prices are hand-sourced and dated (checked 2026-09-16), not scraped. Confirm on the vendor's site before buying.
Why this build
- · A 16 GB RTX 4060 Ti costs a third of a 5090 and still runs 8-13B models comfortably in 4-bit - plenty for a first local setup or a coding-assistant sidecar.
- · AM5 keeps the upgrade path open: the same X670E board and DDR5 kit carry over if you swap in a bigger card later.
- · A 6-core Ryzen 5 is not a bottleneck here - local inference lives on the GPU, so spending more on the CPU buys nothing.
- · 850 W Gold covers this build with room to spare, even after a future GPU upgrade to 300+ W.
Pros
- + Lowest all-new-parts price on the site
- + Same platform as Home AI Lab - a real upgrade path, not a dead end
- + Low power draw, runs on a standard household circuit without a second thought
Limits
- − 16 GB caps you around 13B in 4-bit - no room for 34B+ without swapping the GPU
- − 288 GB/s of memory bandwidth is the slowest card on the site - noticeably slower on long contexts
- − Non-ECC memory, single GPU - not built for training or serving many users
Alternatives
Performance
Home AI Lab: same platform, an RTX 5090 instead - 2x the VRAM for meaningfully more.
Open ↗Balanced
128 GB RAM leaves room for a model library without touching the GPU budget.
Buy vs rent vs hybrid
Over 36 months, at this build's power draw and the live cloud rate for its GPU.
Own it
- Hardware
- $1,748
- Electricity
- $763
- Total cost of ownership
- $2,511
Rent it (cloud)
- Equivalent cloud, monthly
- $1,004/mo
- Equivalent cloud, total
- $36,135
- Break-even
- 2 months
- · Break-even is 2 months, comfortably inside the horizon at this utilization - buying wins.
- · Over 36 months, owning saves roughly $33624 versus the cloud equivalent.
"Buy on Amazon" links are affiliate links - as an Amazon Associate, Obolith earns from qualifying purchases.