Local AI Deals

Cheapest box to run Llama-3.3-70B at Q8_0

Meta’s Llama-3.3-70B is 70B parameters, dense. The number below is computed from that and each vendor’s published specs, not asserted. Switch the quant or the context and it recomputes in place.

Fit finderprices checked 9 hours ago
Quant
Context

Llama-3.3-70B at Q8_0 needs 78GB.

Cheapest Config that fits: DGX Spark, GB10 Grace Blackwell, 128GB. Fits, 42GB headroom. Est. 2.4 tok/s.

Best offer today: $4,699 at NVIDIA in stock, checked 9 hours ago.

Collecting price history — 1 of 90 days. No timing verdict yet.

Offers — DGX Spark, GB10 Grace Blackwell, 128GB, 4TB

  • NVIDIAlist pricein stockchecked 9 hours ago$4,699

GeForce RTX 5090 is a card, not a computer: its price is the US street median plus $1,000 for the minimum host machine. NVIDIA's $1,999 MSRP is not a price anyone pays.

memory = 70B × 1.063 B/param (8.5 bits) + 0.328 GB/1K × 8K K/V + 1.2 GB runtime
speed = bandwidth × 0.65 ÷ (70B active × 1.063 B/param) = 74.4 GB read per token
usable = memory − operating system reserve (8% on a Mac, 6% on a DGX Spark, 4% on a card)

Bandwidth is the vendor’s published figure. Every speed is an estimate and says so. Methodology.

Email me when DGX Spark, GB10 Grace Blackwell, 128GB drops. One email, then we stop.

Pre-filled at the lowest price we have recorded, $4,699.

What else DGX Spark GB10 Grace Blackwell 128GB runs

Best quant that still fits at 8K context, largest model first. Full list on the Config page.

Bandwidth 273GB/s, usable memory 120GB of 128GB.