Cheapest box to run DeepSeek-V4-Flash at BF16
DeepSeek’s DeepSeek-V4-Flash is 284B parameters, 13B 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.
DeepSeek-V4-Flash at BF16 needs 569GB.
Nothing on the Grid runs DeepSeek-V4-Flash at BF16 today. The largest Config, Mac Studio, M5 Ultra, 512GB, is 65GB short. At it fits with 108GB headroom.
Compressed sparse attention. Each layer stores one latent, divided by the compression ratio in its config (mostly 4x or 128x); the figure matches DeepSeek's own claim of about a tenth of the V3 cache.
- Mac mini, M6, 16GB$1,099Won't fit−556GB headroom
- Mac mini, M6, 24GB$1,299Won't fit−548GB headroom
- Mac mini, M6, 32GB$1,499Won't fit−540GB headroom
- Mac mini, M5 Pro, 24GB$1,699Won't fit−548GB headroom
- Mac mini, M5 Pro, 48GB$2,499Won't fit−525GB headroom
- Mac Studio, M5 Max, 36GB$2,499Won't fit−536GB headroom
- Mac mini, M5 Pro, 64GB$2,899Won't fit−510GB headroom
- GeForce RTX 5090, 32GB$2,999.99+hostWon't fit−539GB headroom
- Won't fit−510GB headroom
- Mac Studio, M5 Max, 48GB$3,099Won't fit−525GB headroom
- Mac Studio, M5 Max, 64GB$3,499Won't fit−510GB headroom
- Won't fit−449GB headroom
- Won't fit−481GB headroom
- Won't fit−449GB headroom
- Won't fit−449GB headroom
- Won't fit−449GB headroom
- Won't fit−481GB headroom
- Won't fit−321GB headroom
- Won't fit−321GB headroom
- Won't fit−65GB headroom
| Config | Memory | Price | $/GB | Fit | Headroom | Est. speed |
|---|---|---|---|---|---|---|
| Mac mini, M6 | 16GB | $1,099 | $69 | Won't fit | −556GB | — |
| Mac mini, M6 | 24GB | $1,299 | $54 | Won't fit | −548GB | — |
| Mac mini, M6 | 32GB | $1,499 | $47 | Won't fit | −540GB | — |
| Mac mini, M5 Pro | 24GB | $1,699 | $71 | Won't fit | −548GB | — |
| Mac mini, M5 Pro | 48GB | $2,499 | $52 | Won't fit | −525GB | — |
| Mac Studio, M5 Max | 36GB | $2,499 | $69 | Won't fit | −536GB | — |
| Mac mini, M5 Pro | 64GB | $2,899 | $45 | Won't fit | −510GB | — |
| GeForce RTX 5090 | 32GB | $2,999.99+host | $94 | Won't fit | −539GB | — |
| Mac Studio, M4 Max, refurb | 64GB | $3,049 | $48 | Won't fit | −510GB | — |
| Mac Studio, M5 Max | 48GB | $3,099 | $65 | Won't fit | −525GB | — |
| Mac Studio, M5 Max | 64GB | $3,499 | $55 | Won't fit | −510GB | — |
| Ascent GX10, GB10 Grace Blackwell | 128GB | $3,999.99 | $31 | Won't fit | −449GB | — |
| Mac Studio, M3 Ultra, refurb | 96GB | $4,329 | $45 | Won't fit | −481GB | — |
| Mac Studio, M4 Max, refurb | 128GB | $4,409 | $34 | Won't fit | −449GB | — |
| DGX Spark, GB10 Grace Blackwell | 128GB | $4,699 | $37 | Won't fit | −449GB | — |
| Mac Studio, M5 Max | 128GB | $5,099 | $40 | Won't fit | −449GB | — |
| Mac Studio, M5 Ultra | 96GB | $5,499 | $57 | Won't fit | −481GB | — |
| Mac Studio, M3 Ultra, refurb | 256GB | $7,729 | $30 | Won't fit | −321GB | — |
| Mac Studio, M5 Ultra | 256GB | $9,499 | $37 | Won't fit | −321GB | — |
| Mac Studio, M5 Ultra | 512GB | — | — | Won't fit | −65GB | — |
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 = 284B × 2.000 B/param (16 bits) + 0.006 GB/1K × 8K K/V + 1.2 GB runtime
speed = bandwidth × 0.5 ÷ (13B active × 2.000 B/param) = 26.0 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.