Cheapest box to run Llama-4-Scout-17B-16E at BF16
Meta’s Llama-4-Scout-17B-16E is 109B parameters, 17B 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.
Llama-4-Scout-17B-16E at BF16 needs 221GB.
Cheapest Config that fits: Mac Studio, M5 Ultra, 256GB. Fits, 27GB headroom. Est. 17.6 tok/s.
Best offer today: $9,499 at Apple — in stock, checked 10 hours ago.
Collecting price history — 2 of 90 days. No timing verdict yet.
Offers — Mac Studio, M5 Ultra, 256GB, 1TB
- Mac mini, M6, 16GB$1,099Won't fit−208GB headroom
- Mac mini, M6, 24GB$1,299Won't fit−200GB headroom
- Mac mini, M6, 32GB$1,499Won't fit−192GB headroom
- Mac mini, M5 Pro, 24GB$1,699Won't fit−200GB headroom
- Mac mini, M5 Pro, 48GB$2,499Won't fit−177GB headroom
- Mac Studio, M5 Max, 36GB$2,499Won't fit−188GB headroom
- Mac mini, M5 Pro, 64GB$2,899Won't fit−162GB headroom
- GeForce RTX 5090, 32GB$2,999.99+hostWon't fit−190GB headroom
- Mac Studio, M5 Max, 48GB$3,099Won't fit−177GB headroom
- Mac Studio, M5 Max, 64GB$3,499Won't fit−162GB headroom
- Won't fit−100GB headroom
- Won't fit−100GB headroom
- Won't fit−101GB headroom
- Won't fit−132GB headroom
- Fits27GB headroomest. 17.6 tok/s
- Fits283GB headroomest. 17.6 tok/s
| Config | Memory | Price | $/GB | Fit | Headroom | Est. speed |
|---|---|---|---|---|---|---|
| Mac mini, M6 | 16GB | $1,099 | $69 | Won't fit | −208GB | — |
| Mac mini, M6 | 24GB | $1,299 | $54 | Won't fit | −200GB | — |
| Mac mini, M6 | 32GB | $1,499 | $47 | Won't fit | −192GB | — |
| Mac mini, M5 Pro | 24GB | $1,699 | $71 | Won't fit | −200GB | — |
| Mac mini, M5 Pro | 48GB | $2,499 | $52 | Won't fit | −177GB | — |
| Mac Studio, M5 Max | 36GB | $2,499 | $69 | Won't fit | −188GB | — |
| Mac mini, M5 Pro | 64GB | $2,899 | $45 | Won't fit | −162GB | — |
| GeForce RTX 5090 | 32GB | $2,999.99+host | $94 | Won't fit | −190GB | — |
| Mac Studio, M5 Max | 48GB | $3,099 | $65 | Won't fit | −177GB | — |
| Mac Studio, M5 Max | 64GB | $3,499 | $55 | Won't fit | −162GB | — |
| Ascent GX10, GB10 Grace Blackwell | 128GB | $3,999.99 | $31 | Won't fit | −100GB | — |
| DGX Spark, GB10 Grace Blackwell | 128GB | $4,699 | $37 | Won't fit | −100GB | — |
| Mac Studio, M5 Max | 128GB | $5,099 | $40 | Won't fit | −101GB | — |
| Mac Studio, M5 Ultra | 96GB | $5,499 | $57 | Won't fit | −132GB | — |
| Mac Studio, M5 Ultra | 256GB | $9,499 | $37 | Fits | 27GB | 17.6 tok/s |
| Mac Studio, M5 Ultra | 512GB | — | — | Fits | 283GB | 17.6 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 = 109B × 2.000 B/param (16 bits) + 0.197 GB/1K × 8K K/V + 1.2 GB runtime
speed = bandwidth × 0.5 ÷ (17B active × 2.000 B/param) = 34.0 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 Mac Studio M5 Ultra 256GB runs
- Llama-4-Maverick-17B-128EQ4_K_M, est. 58.5 tok/s
- Qwen3-235B-A22BQ6_K, est. 33.3 tok/s
- Mistral-Large-2411Q8_0, est. 6.0 tok/s
- gpt-oss-120bBF16, est. 58.8 tok/s
- Llama-4-Scout-17B-16EBF16, est. 17.6 tok/s
- Llama-3.3-70BBF16, est. 5.6 tok/s
Best quant that still fits at 8K context, largest model first. Full list on the Config page.
Bandwidth 1.2TB/s, usable memory 248GB of 256GB.