Cheapest box to run Qwen3-32B at BF16
Alibaba’s Qwen3-32B is 32B 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.
Qwen3-32B at BF16 needs 67GB.
Cheapest Config that fits: Ascent GX10, GB10 Grace Blackwell, 128GB. Fits, 53GB headroom. Est. 2.8 tok/s.
Best offer today: $3,999.99 at Best Buy — in stock, checked 10 hours ago.
Collecting price history — 2 of 90 days. No timing verdict yet.
Offers — Ascent GX10, GB10 Grace Blackwell, 128GB, 1TB
- Mac mini, M6, 16GB$1,099Won't fit−54GB headroom
- Mac mini, M6, 24GB$1,299Won't fit−46GB headroom
- Mac mini, M6, 32GB$1,499Won't fit−38GB headroom
- Mac mini, M5 Pro, 24GB$1,699Won't fit−46GB headroom
- Mac mini, M5 Pro, 48GB$2,499Won't fit−23GB headroom
- Mac Studio, M5 Max, 36GB$2,499Won't fit−34GB headroom
- Mac mini, M5 Pro, 64GB$2,899Won't fit−8GB headroom
- GeForce RTX 5090, 32GB$2,999.99+hostWon't fit−37GB headroom
- Mac Studio, M5 Max, 48GB$3,099Won't fit−23GB headroom
- Mac Studio, M5 Max, 64GB$3,499Won't fit−8GB headroom
- Fits53GB headroomest. 2.8 tok/s
- Fits53GB headroomest. 2.8 tok/s
- Fits53GB headroomest. 6.2 tok/s
- Fits21GB headroomest. 12.2 tok/s
- Fits181GB headroomest. 12.2 tok/s
- Fits437GB headroomest. 12.2 tok/s
| Config | Memory | Price | $/GB | Fit | Headroom | Est. speed |
|---|---|---|---|---|---|---|
| Mac mini, M6 | 16GB | $1,099 | $69 | Won't fit | −54GB | — |
| Mac mini, M6 | 24GB | $1,299 | $54 | Won't fit | −46GB | — |
| Mac mini, M6 | 32GB | $1,499 | $47 | Won't fit | −38GB | — |
| Mac mini, M5 Pro | 24GB | $1,699 | $71 | Won't fit | −46GB | — |
| Mac mini, M5 Pro | 48GB | $2,499 | $52 | Won't fit | −23GB | — |
| Mac Studio, M5 Max | 36GB | $2,499 | $69 | Won't fit | −34GB | — |
| Mac mini, M5 Pro | 64GB | $2,899 | $45 | Won't fit | −8GB | — |
| GeForce RTX 5090 | 32GB | $2,999.99+host | $94 | Won't fit | −37GB | — |
| Mac Studio, M5 Max | 48GB | $3,099 | $65 | Won't fit | −23GB | — |
| Mac Studio, M5 Max | 64GB | $3,499 | $55 | Won't fit | −8GB | — |
| Ascent GX10, GB10 Grace Blackwell | 128GB | $3,999.99 | $31 | Fits | 53GB | 2.8 tok/s |
| DGX Spark, GB10 Grace Blackwell | 128GB | $4,699 | $37 | Fits | 53GB | 2.8 tok/s |
| Mac Studio, M5 Max | 128GB | $5,099 | $40 | Fits | 53GB | 6.2 tok/s |
| Mac Studio, M5 Ultra | 96GB | $5,499 | $57 | Fits | 21GB | 12.2 tok/s |
| Mac Studio, M5 Ultra | 256GB | $9,499 | $37 | Fits | 181GB | 12.2 tok/s |
| Mac Studio, M5 Ultra | 512GB | — | — | Fits | 437GB | 12.2 tok/s |
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 = 32B × 2.000 B/param (16 bits) + 0.262 GB/1K × 8K K/V + 1.2 GB runtime
speed = bandwidth × 0.65 ÷ (32B active × 2.000 B/param) = 64.0 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.
What else Ascent GX10 GB10 Grace Blackwell 128GB runs
- Qwen3-235B-A22BQ3_K_M, est. 12.7 tok/s
- Mistral-Large-2411Q6_K, est. 1.8 tok/s
- gpt-oss-120bQ6_K, est. 32.6 tok/s
- Llama-4-Scout-17B-16EQ8_0, est. 7.6 tok/s
- Llama-3.3-70BQ8_0, est. 2.4 tok/s
- Qwen3-32BBF16, est. 2.8 tok/s
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.