Cheapest box to run Qwen3-Coder-Next at Q4_K_M
Alibaba’s Qwen3-Coder-Next is 80B parameters, 3B 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.
Qwen3-Coder-Next at Q4_K_M needs 50GB.
Cheapest Config that fits: Mac mini, M5 Pro, 64GB. Fits, 9GB headroom. Est. 84.7 tok/s.
Best offer today: $2,899 at Apple — in stock, checked 50 minutes ago.
Price history: 34 days recorded.
The Qwen3-Next hybrid again: 12 of 48 layers are full attention and the linear layers cache nothing per token.
Offers — Mac mini, M5 Pro, 64GB, 512GB
- Mac mini, M6, 16GB$1,099Won't fit−37GB headroom
- Mac mini, M6, 24GB$1,299Won't fit−29GB headroom
- Mac mini, M6, 32GB$1,499Won't fit−21GB headroom
- Mac mini, M5 Pro, 24GB$1,699Won't fit−29GB headroom
- Mac mini, M5 Pro, 48GB$2,499Won't fit−6GB headroom
- Mac Studio, M5 Max, 36GB$2,499Won't fit−17GB headroom
- Mac mini, M5 Pro, 64GB$2,899Fits9GB headroomest. 84.7 tok/s
- GeForce RTX 5090, 32GB$2,999.99+hostWon't fit−19GB headroom
- Fits9GB headroomest. 151 tok/s
- Mac Studio, M5 Max, 48GB$3,099Won't fit−6GB headroom
- Mac Studio, M5 Max, 64GB$3,499Fits9GB headroomest. 169 tok/s
- Fits71GB headroomest. 75.4 tok/s
- Fits39GB headroomest. 226 tok/s
- Fits70GB headroomest. 151 tok/s
- Fits71GB headroomest. 75.4 tok/s
- Fits70GB headroomest. 169 tok/s
- Fits39GB headroomest. 331 tok/s
- Fits198GB headroomest. 226 tok/s
- Fits198GB headroomest. 331 tok/s
- Fits454GB headroomest. 331 tok/s
| Config | Memory | Price | $/GB | Fit | Headroom | Est. speed |
|---|---|---|---|---|---|---|
| Mac mini, M6 | 16GB | $1,099 | $69 | Won't fit | −37GB | — |
| Mac mini, M6 | 24GB | $1,299 | $54 | Won't fit | −29GB | — |
| Mac mini, M6 | 32GB | $1,499 | $47 | Won't fit | −21GB | — |
| Mac mini, M5 Pro | 24GB | $1,699 | $71 | Won't fit | −29GB | — |
| Mac mini, M5 Pro | 48GB | $2,499 | $52 | Won't fit | −6GB | — |
| Mac Studio, M5 Max | 36GB | $2,499 | $69 | Won't fit | −17GB | — |
| Mac mini, M5 Pro | 64GB | $2,899 | $45 | Fits | 9GB | 84.7 tok/s |
| GeForce RTX 5090 | 32GB | $2,999.99+host | $94 | Won't fit | −19GB | — |
| Mac Studio, M4 Max, refurb | 64GB | $3,049 | $48 | Fits | 9GB | 151 tok/s |
| Mac Studio, M5 Max | 48GB | $3,099 | $65 | Won't fit | −6GB | — |
| Mac Studio, M5 Max | 64GB | $3,499 | $55 | Fits | 9GB | 169 tok/s |
| Ascent GX10, GB10 Grace Blackwell | 128GB | $3,999.99 | $31 | Fits | 71GB | 75.4 tok/s |
| Mac Studio, M3 Ultra, refurb | 96GB | $4,329 | $45 | Fits | 39GB | 226 tok/s |
| Mac Studio, M4 Max, refurb | 128GB | $4,409 | $34 | Fits | 70GB | 151 tok/s |
| DGX Spark, GB10 Grace Blackwell | 128GB | $4,699 | $37 | Fits | 71GB | 75.4 tok/s |
| Mac Studio, M5 Max | 128GB | $5,099 | $40 | Fits | 70GB | 169 tok/s |
| Mac Studio, M5 Ultra | 96GB | $5,499 | $57 | Fits | 39GB | 331 tok/s |
| Mac Studio, M3 Ultra, refurb | 256GB | $7,729 | $30 | Fits | 198GB | 226 tok/s |
| Mac Studio, M5 Ultra | 256GB | $9,499 | $37 | Fits | 198GB | 331 tok/s |
| Mac Studio, M5 Ultra | 512GB | — | — | Fits | 454GB | 331 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 = 80B × 0.604 B/param (4.83 bits) + 0.025 GB/1K × 8K K/V + 1.2 GB runtime
speed = bandwidth × 0.5 ÷ (3B active × 0.604 B/param) = 1.8 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 mini M5 Pro 64GB runs
- Llama-4-Scout-17B-16EQ3_K_M, est. 18.5 tok/s
- Qwen3-Coder-NextQ5_K_M, est. 72.2 tok/s
- Llama-3.3-70BQ5_K_M, est. 4.0 tok/s
- Qwen3.5-35B-A3BQ8_0, est. 48.2 tok/s
- Qwen3-32BQ8_0, est. 5.9 tok/s
- Gemma-4-31BQ8_0, est. 6.1 tok/s
Best quant that still fits at 8K context, largest model first. Full list on the Config page.
Bandwidth 307GB/s, usable memory 59GB of 64GB.