Best budget local AI machine (under $3,000)
Our pick for under $3,000 is the GMKtec EVO-X2 64GB / 1TB: $2,200 / €2,000, 94.6 tok/s on Qwen3-Coder-30B-A3B, which works out to 43.0 tok/s per $1,000. Runner-up: Framework Desktop 64GB (DIY) at ≈$2,486 / €2,209 and 94.6 tok/s.
No complete machine we track is under $2,000 right now. Machines under $3,000 have 64GB, which runs mid-size models like Qwen3-Coder-30B-A3B but not gpt-oss-120b.
Buy the pick: Amazon US · Amazon EU
| # | Machine | Price | tok/s | per $1k |
|---|---|---|---|---|
| 1 | GMKtec EVO-X2 64GB / 1TB Strix Halo · 64GB · checked 2026-10-16 | $2,200 €2,000 | 94.6 | 43.0 |
| 2 | Framework Desktop 64GB (DIY) Strix Halo · 64GB · checked 2026-10-16 | ≈$2,486 €2,209 | 94.6 | 38.1 |
| 3 | Minisforum MS-S1 MAX 64GB / 2TB Strix Halo · 64GB · checked 2026-10-16 | $2,599 ≈€2,310 | 94.6 | 36.4 |
tok/s = generation speed on Qwen3-Coder-30B-A3B from public benchmarks. per $1k = tok/s per $1,000 of the US price. ≈ = converted at the ECB rate.
Not ranked: Mac mini M5 Pro 16-core GPU (no public result yet); Mac mini M5 Pro 20-core GPU (no public result yet).
How we picked
Ranked by measured tokens per second per $1,000 of the current US price. Speeds are public benchmark results, one per chip, linked on each product page and on Benchmarks; engines and quants differ between sources. Prices checked 2026-10-16 19:02 UTC and refreshed every 6 hours.
FAQ
What is the cheapest computer for running local LLMs?
Among complete machines with public results, GMKtec EVO-X2 64GB / 1TB at $2,200 / €2,000.
Is there a good local AI machine under $2,000?
Not among the complete machines we track right now.
Can a 64GB machine run gpt-oss-120b?
No, it needs about 64GB for the weights alone. Look at the 128GB page instead.