Cheapest box to run Kimi-K2-Instruct at Q4_K_M
Moonshot’s Kimi-K2-Instruct is 1000B parameters, 32B 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.
Kimi-K2-Instruct at Q4_K_M needs 606GB.
Nothing on the Grid runs Kimi-K2-Instruct at Q4_K_M today. The largest Config, Mac Studio, M5 Ultra, 512GB, is 102GB short. At it fits with 13GB headroom.
A trillion parameters. No Config we track holds it at any quant, and we list it so the arithmetic says so plainly.
- Mac mini, M6, 16GB$1,099Won't fit−593GB headroom
- Mac mini, M6, 24GB$1,299Won't fit−585GB headroom
- Mac mini, M6, 32GB$1,499Won't fit−577GB headroom
- Mac mini, M5 Pro, 24GB$1,699Won't fit−585GB headroom
- Mac mini, M5 Pro, 48GB$2,499Won't fit−561GB headroom
- Mac Studio, M5 Max, 36GB$2,499Won't fit−573GB headroom
- Mac mini, M5 Pro, 64GB$2,899Won't fit−547GB headroom
- GeForce RTX 5090, 32GB$2,999.99+hostWon't fit−575GB headroom
- Won't fit−547GB headroom
- Mac Studio, M5 Max, 48GB$3,099Won't fit−561GB headroom
- Mac Studio, M5 Max, 64GB$3,499Won't fit−547GB headroom
- Won't fit−485GB headroom
- Won't fit−517GB headroom
- Won't fit−486GB headroom
- Won't fit−485GB headroom
- Won't fit−486GB headroom
- Won't fit−517GB headroom
- Won't fit−358GB headroom
- Won't fit−358GB headroom
- Won't fit−102GB headroom
| Config | Memory | Price | $/GB | Fit | Headroom | Est. speed |
|---|---|---|---|---|---|---|
| Mac mini, M6 | 16GB | $1,099 | $69 | Won't fit | −593GB | — |
| Mac mini, M6 | 24GB | $1,299 | $54 | Won't fit | −585GB | — |
| Mac mini, M6 | 32GB | $1,499 | $47 | Won't fit | −577GB | — |
| Mac mini, M5 Pro | 24GB | $1,699 | $71 | Won't fit | −585GB | — |
| Mac mini, M5 Pro | 48GB | $2,499 | $52 | Won't fit | −561GB | — |
| Mac Studio, M5 Max | 36GB | $2,499 | $69 | Won't fit | −573GB | — |
| Mac mini, M5 Pro | 64GB | $2,899 | $45 | Won't fit | −547GB | — |
| GeForce RTX 5090 | 32GB | $2,999.99+host | $94 | Won't fit | −575GB | — |
| Mac Studio, M4 Max, refurb | 64GB | $3,049 | $48 | Won't fit | −547GB | — |
| Mac Studio, M5 Max | 48GB | $3,099 | $65 | Won't fit | −561GB | — |
| Mac Studio, M5 Max | 64GB | $3,499 | $55 | Won't fit | −547GB | — |
| Ascent GX10, GB10 Grace Blackwell | 128GB | $3,999.99 | $31 | Won't fit | −485GB | — |
| Mac Studio, M3 Ultra, refurb | 96GB | $4,329 | $45 | Won't fit | −517GB | — |
| Mac Studio, M4 Max, refurb | 128GB | $4,409 | $34 | Won't fit | −486GB | — |
| DGX Spark, GB10 Grace Blackwell | 128GB | $4,699 | $37 | Won't fit | −485GB | — |
| Mac Studio, M5 Max | 128GB | $5,099 | $40 | Won't fit | −486GB | — |
| Mac Studio, M5 Ultra | 96GB | $5,499 | $57 | Won't fit | −517GB | — |
| Mac Studio, M3 Ultra, refurb | 256GB | $7,729 | $30 | Won't fit | −358GB | — |
| Mac Studio, M5 Ultra | 256GB | $9,499 | $37 | Won't fit | −358GB | — |
| Mac Studio, M5 Ultra | 512GB | — | — | Won't fit | −102GB | — |
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 = 1000B × 0.604 B/param (4.83 bits) + 0.070 GB/1K × 8K K/V + 1.2 GB runtime
speed = bandwidth × 0.5 ÷ (32B active × 0.604 B/param) = 19.3 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.