Cheapest box to run Llama-4-Maverick-17B-128E at BF16
Meta’s Llama-4-Maverick-17B-128E is 400B 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-Maverick-17B-128E at BF16 needs 803GB.
Nothing on the Grid runs Llama-4-Maverick-17B-128E at BF16 today. The largest Config, Mac Studio, M5 Ultra, 512GB, is 299GB short. At it fits with 50GB headroom.
- Mac mini, M6, 16GB$1,099Won't fit−790GB headroom
- Mac mini, M6, 24GB$1,299Won't fit−782GB headroom
- Mac mini, M6, 32GB$1,499Won't fit−774GB headroom
- Mac mini, M5 Pro, 24GB$1,699Won't fit−782GB headroom
- Mac mini, M5 Pro, 48GB$2,499Won't fit−759GB headroom
- Mac Studio, M5 Max, 36GB$2,499Won't fit−770GB headroom
- Mac mini, M5 Pro, 64GB$2,899Won't fit−744GB headroom
- GeForce RTX 5090, 32GB$2,999.99+hostWon't fit−772GB headroom
- Mac Studio, M5 Max, 48GB$3,099Won't fit−759GB headroom
- Mac Studio, M5 Max, 64GB$3,499Won't fit−744GB headroom
- Won't fit−682GB headroom
- Won't fit−682GB headroom
- Won't fit−683GB headroom
- Won't fit−714GB headroom
- Won't fit−555GB headroom
- Won't fit−299GB headroom
| Config | Memory | Price | $/GB | Fit | Headroom | Est. speed |
|---|---|---|---|---|---|---|
| Mac mini, M6 | 16GB | $1,099 | $69 | Won't fit | −790GB | — |
| Mac mini, M6 | 24GB | $1,299 | $54 | Won't fit | −782GB | — |
| Mac mini, M6 | 32GB | $1,499 | $47 | Won't fit | −774GB | — |
| Mac mini, M5 Pro | 24GB | $1,699 | $71 | Won't fit | −782GB | — |
| Mac mini, M5 Pro | 48GB | $2,499 | $52 | Won't fit | −759GB | — |
| Mac Studio, M5 Max | 36GB | $2,499 | $69 | Won't fit | −770GB | — |
| Mac mini, M5 Pro | 64GB | $2,899 | $45 | Won't fit | −744GB | — |
| GeForce RTX 5090 | 32GB | $2,999.99+host | $94 | Won't fit | −772GB | — |
| Mac Studio, M5 Max | 48GB | $3,099 | $65 | Won't fit | −759GB | — |
| Mac Studio, M5 Max | 64GB | $3,499 | $55 | Won't fit | −744GB | — |
| Ascent GX10, GB10 Grace Blackwell | 128GB | $3,999.99 | $31 | Won't fit | −682GB | — |
| DGX Spark, GB10 Grace Blackwell | 128GB | $4,699 | $37 | Won't fit | −682GB | — |
| Mac Studio, M5 Max | 128GB | $5,099 | $40 | Won't fit | −683GB | — |
| Mac Studio, M5 Ultra | 96GB | $5,499 | $57 | Won't fit | −714GB | — |
| Mac Studio, M5 Ultra | 256GB | $9,499 | $37 | Won't fit | −555GB | — |
| Mac Studio, M5 Ultra | 512GB | — | — | Won't fit | −299GB | — |
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 = 400B × 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.