landing page

Local LLM under $1,000

Global prices (USD), EUR available. Under $1,000 you buy an honest 8B or 14B machine. A 70B that fully fits sits just above that in this dataset, or just under it and slow.

Under €1000 this list lands on a used 16 GB M4 Mac mini, a 16 GB M6 Mac mini, or a modest PC around an RTX 3060 12GB. That runs 8B well enough to work with, and 14B Q4 just barely on the 3060.

32B Q4 wants a bit over 20 GB. That fits on a 3090, not on 12 or 16 GB, and that 3090 PC sits above €1000 in our estimate. 70B does not belong here if you want the model fully in memory.

The page below lets the engine pick a model itself: the largest one that fully fits inside the budget, at Q4 and 8,192 tokens.

Open in the calculator

Three setups

Phi-4 14B · Q4 · 8,192 ctx

Estimated memory

10.7 GBestimate

8.2 GB weights · 1.7 GB KV cache · 0.8 GB overhead

You didn't pick a model. Inside this budget, this is the largest model that fully fits somewhere.

01 · Fastest
Budget PC with RTX 3060 12GB
Entry point for 8B and, just barely, 14B Q4. The card is mostly second-hand.
$996€890
14 Oct 2026
price unverified
inside budget · used
tok/s
19.0
estimate
needed
10.7 GB
full fit
usable
10.8 GB
VRAM
Why this one
- The model fits entirely in VRAM or unified memory. No CPU offload.
- It fits, with little room left. A longer context can push it over.
- The indicative price is inside your budget.
- Second-hand or a previous generation. The price moves around.
- Highest estimated speed in this shortlist.
Parts
- GeForce RTX 3060 12GB
- Ryzen 5 5600 or similar
- 32 GB DDR4
- 1 TB SSD, PSU around 550W
EU shops
US shops are not linked yet.
02 · A step up
PC with RTX 5090 32GB
Highest bandwidth in a single card. 32 GB is plenty for 32B, still too tight for 70B Q4.
$8,926€7,973
15 Oct 2026
price unverified
Not a checked system price. The GPU line is the €6,273 PriceSquirrel figure from 15 Oct 2026. The rest of the box was not rechecked.
over budget
tok/s
80
estimate
needed
10.7 GB
full fit
usable
30.8 GB
VRAM
Why this one
- The model fits entirely in VRAM or unified memory. No CPU offload.
- Memory left for a longer context or a heavier quantisation.
- Above your budget. Nothing fully fit inside the amount, or this is a faster step above it.
- High memory bandwidth, so a higher tokens-per-second estimate.
Parts
- GeForce RTX 5090 32GB
- Ryzen 7 9700X or similar
- 96 GB DDR5
- 1200W PSU, roomy case
EU shops
US shops are not linked yet.
03 · A step up
Dual RTX 4090, 48 GB
Same memory as two 3090s, higher bandwidth, a lot more money.
$6,146€5,490
14 Oct 2026
price unverified
over budget
tok/s
73
estimate
needed
10.7 GB
full fit
usable
45.6 GB
VRAM
Why this one
- The model fits entirely in VRAM or unified memory. No CPU offload.
- Memory left for a longer context or a heavier quantisation.
- Above your budget. Nothing fully fit inside the amount, or this is a faster step above it.
- Two GPUs. Expect a heavier PSU and a llama.cpp split. Bandwidth does not fully double.
- High memory bandwidth, so a higher tokens-per-second estimate.
Parts
- 2× GeForce RTX 4090 24GB
- Ryzen 9 7950X or similar
- 128 GB DDR5
- 1600W PSU
EU shops
US shops are not linked yet.

Tokens per second are estimated from memory bandwidth, not from a measured benchmark. EUR prices include VAT. USD is an FX estimate. A price without a checked listing is marked price unverified. Run llama-bench yourself before you buy.

Want a build tailored to you? Custom report

Local LLM under $1,000

Three setups

Budget PC with RTX 3060 12GB

PC with RTX 5090 32GB

Dual RTX 4090, 48 GB

Get new builds when GPU prices move