Local AI Deals

Cheapest box to run Qwen3-Coder-Next at BF16

Alibaba’s Qwen3-Coder-Next is 80B parameters, 3B of them read per token. 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 2 hours ago
Quant
Context

Qwen3-Coder-Next at BF16 needs 161GB.

Cheapest Config that fits: Mac Studio, M3 Ultra, 256GB, refurbished. Fits, 87GB headroom. Est. 68.3 tok/s.

Best offer today: $7,729 at Apple Certified Refurbished — in stock, checked 2 hours ago.

Price history: 34 days recorded.

The Qwen3-Next hybrid again: 12 of 48 layers are full attention and the linear layers cache nothing per token.

Offers — Mac Studio, M3 Ultra, 256GB, 1TB, refurbished

  • Apple Certified RefurbishedRefurbished by Apple, with Apple's one-year limited warranty and 14-day returns. Apple says refurbished supply is usually very limited and often runs out, so this row can go dead faster than a new one.in stockchecked 2 hours ago$7,729

GeForce RTX 5090 is a card, not a computer: every price shown for it is the cheapest listing we found plus $1,000 for the minimum host machine.

memory = 80B × 2.000 B/param (16 bits) + 0.025 GB/1K × 8K K/V + 1.2 GB runtime
speed = bandwidth × 0.5 ÷ (3B active × 2.000 B/param) = 6.0 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 Mac Studio, M3 Ultra, 256GB, refurbished drops. One email, then we stop.

$7,729 is the lowest price we have recorded, rounded down to the dollar.

What else Mac Studio M3 Ultra 256GB refurb runs

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

Bandwidth 819GB/s, usable memory 248GB of 256GB.