RigPrice.

Buy vs rent

Given a model and your monthly usage, when does buying used hardware beat renting a cloud GPU? The verdict comes only from the formula below, and both paths get equal billing. Hardware costs use listings that passed our trust screen (screened, not guaranteed).

Popular models at default assumptions

Defaults: 40 h/mo · $0.16/kWh · 24-month horizon · resale at 65% of median ask.

Qwen3.6 27B (Q4_K_M, needs 20.88 GB VRAM)

Renting wins at 40 h/mo (breakeven ≈ 154 months, beyond the 24-month horizon).

Buy hardware

$898

1× AMD Radeon RX 7900 XTX: 24 GB VRAM, ~430 W system

Est. resale
− $605
Net hardware cost
$293
Power at $0.16/kWh
$0.069/hr

used · 2,597 feedback · 99.6% positive · Top Rated seller · returns accepted · listed 3d

Screened, not guaranteed. Eligible purchases are covered by the eBay Money Back Guarantee.

Rent cloud

$0.116/hr

Vast.ai: cheapest offer fitting 20.88 GB (rtx-3090 class, on-demand)

40 h/mo
$4.65/mo
Total over 24 months
$112

Cloud prices checked 2026-08-02. Storage/egress excluded.

Qwen3.6 35B-A3B (Q4_K_M, needs 25.41 GB VRAM)

Renting wins at 40 h/mo (breakeven ≈ 28 months, beyond the 24-month horizon).

Buy hardware

$898

1× AMD Radeon RX 7900 XTX: 24 GB VRAM, ~430 W system

Est. resale
− $605
Net hardware cost
$293
Power at $0.16/kWh
$0.069/hr

used · 2,597 feedback · 99.6% positive · Top Rated seller · returns accepted · listed 3d

Screened, not guaranteed. Eligible purchases are covered by the eBay Money Back Guarantee.

Rent cloud

$0.33/hr

RunPod: cheapest offer fitting 25.41 GB (rtx-a6000 class, on-demand)

40 h/mo
$13.2/mo
Total over 24 months
$317

Cloud prices checked 2026-08-02. Storage/egress excluded.

Gemma 4 31B (Q4_K_M, needs 21.22 GB VRAM)

Renting wins at 40 h/mo (breakeven ≈ 154 months, beyond the 24-month horizon).

Buy hardware

$898

1× AMD Radeon RX 7900 XTX: 24 GB VRAM, ~430 W system

Est. resale
− $605
Net hardware cost
$293
Power at $0.16/kWh
$0.069/hr

used · 2,597 feedback · 99.6% positive · Top Rated seller · returns accepted · listed 3d

Screened, not guaranteed. Eligible purchases are covered by the eBay Money Back Guarantee.

Rent cloud

$0.116/hr

Vast.ai: cheapest offer fitting 21.22 GB (rtx-3090 class, on-demand)

40 h/mo
$4.65/mo
Total over 24 months
$112

Cloud prices checked 2026-08-02. Storage/egress excluded.

Assumptions behind these numbers (from the data export, verbatim)
assumptions.resale_fraction
0.65
assumptions.kwh_price
0.16
assumptions.hours_per_month
40
assumptions.horizon_months
24
assumptions.power_model
power_per_hr = est_system_watts / 1000 * kwh_price
assumptions.excluded_costs
cloud storage/egress excluded
assumptions.hw_cost_basis
sum of the gpu_count cheapest active buy_pick listings (used bucket); a config whose GPU has fewer screened listings than it needs is left unpriced
assumptions.resale_basis
resale_fraction x current median_ask x gpu_count
assumptions.bandwidth_utilization
0.5
assumptions.speed_model
est_toks_per_s = mem_bandwidth_gbps x bandwidth_utilization / (file_gb x active_params_b / params_b); per-card bandwidth (layer-split decode is sequential), spec-sheet numbers, estimate not benchmark
assumptions.recommend_rule
three personas from the first priced quant, caveat/cooling-mod cards never win: cheapest = lowest cost clean config (tight fits and multi-GPU allowed); easiest = cheapest comfortable single clean card at >= 10 est tok/s, NVIDIA preferred, with fastest/multi-GPU fallbacks; fastest = highest est tok/s among comfortable fits, ties to cheaper
assumptions.tight_fit_basis
min_vram_tight_gb = file_gb + ~1.5 GB (about 4k context with a quantized KV cache); configs between the tight and comfortable bars publish as tight_fit