budget
Local LLM under $3,000
Under $3,000 (€2,680 incl. VAT) localLLMrig picks Qwen2.5 72B at Q4 as the largest model that fully fits somewhere. Global prices (USD), EUR available. Those prices are marked price unverified. Tokens/s are estimates.
This page is $3,000. The catalog compares against about €2,680 incl. VAT, an FX estimate.
Three setups
Qwen2.5 72B · Q4 · 8,192 ctx
Estimated memory
44.2 GBestimate
40.7 GB weights · 2.7 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
Dual RTX 3090, 48 GB
The classic 70B Q4 rig. Two cards, more hassle, enough memory.
$2,217€1,980
14 Oct 2026
price unverifiedinside budget · used
tok/s
18.1estimate
needed
44.2 GBfull fit
usable
45.6 GBVRAM
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.
- 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.
- Second-hand or a previous generation. The price moves around.
- Highest estimated speed in this shortlist.
Parts
- 2× GeForce RTX 3090 24GB, used
- Ryzen 9 5900X or similar
- 128 GB DDR4
- 1200W PSU, a case that can breathe
EU shops
US shops are not linked yet.
02 · Other platform
Mac mini M4 Pro 64GB
Enough memory for 70B Q4, but bandwidth is modest. It fits; it is not fast.
$2,116€1,890
14 Oct 2026
price unverifiedinside budget · used
tok/s
3.0estimate
needed
44.2 GBfull fit
usable
54.0 GBunified
Why this one
- The model fits entirely in VRAM or unified memory. No CPU offload.
- The indicative price is inside your budget.
- Unified memory: weights and KV cache share one pool.
- Second-hand or a previous generation. The price moves around.
- One box, little noise, no discrete GPU.
- A different brand, so you don't see the same PC three times.
Parts
- Mac mini M4 Pro, 64 GB unified memory
- 273 GB/s bandwidth is an estimate; no page was re-fetched
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 unverifiedover budget
tok/s
19.4estimate
needed
44.2 GBfull fit
usable
45.6 GBVRAM
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.
- 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.
With CPU offload, and therefore slower, this could also work: PC with Radeon RX 7900 XTX 24GB, ~1.8 tok/s, $2,004€1,790.
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
Get new builds when GPU prices move
End of results
Common questions
- What is the largest model that fully fits under $3,000?
- In this dataset that is Qwen2.5 72B at Q4, with an estimated memory need of 44.2 GB at 8,192 tokens. That is an estimate. sourceUrl stays null.
- Is a 70B model always included?
- No. Llama 70B Q4 needs about 40 GB. At the lower amounts it does not fully fit. The page then shows a smaller model that does, or a fit above the budget when nothing inside it does.
- Are the prices shop prices?
- The setups on this page are marked price unverified. EUR includes VAT. USD is an FX estimate. The RTX 5090 card is the catalog row with a checked listing, and it sits above these budgets.