Cheapest box to run Qwen3-235B-A22B at Q3_K_M
Alibaba’s Qwen3-235B-A22B is 235B parameters, 22B 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.
Qwen3-235B-A22B at Q3_K_M needs 118GB.
Cheapest Config that fits: Mac Studio, M5 Ultra, 256GB. Fits, 130GB headroom. Est. 55.8 tok/s.
Best offer today: $9,499 at Apple — in stock, checked 6 hours ago.
Collecting price history — 1 of 90 days. No timing verdict yet.
Offers — Mac Studio, M5 Ultra, 256GB, 1TB
- Mac mini, M6, 16GB$1,099Won't fit−105GB headroom
- Mac mini, M6, 24GB$1,299Won't fit−97GB headroom
- Mac mini, M6, 32GB$1,499Won't fit−89GB headroom
- Mac mini, M5 Pro, 24GB$1,699Won't fit−97GB headroom
- Mac mini, M5 Pro, 48GB$2,499Won't fit−73GB headroom
- Mac Studio, M5 Max, 36GB$2,499Won't fit−85GB headroom
- Mac mini, M5 Pro, 64GB$2,899Won't fit−59GB headroom
- GeForce RTX 5090, 32GB$2,999.99+hostWon't fit−87GB headroom
- Mac Studio, M5 Max, 48GB$3,099Won't fit−73GB headroom
- Mac Studio, M5 Max, 64GB$3,499Won't fit−59GB headroom
- Tight3GB headroomest. 12.7 tok/s
- Tight3GB headroomest. 12.7 tok/s
- Tight2GB headroomest. 28.6 tok/s
- Won't fit−29GB headroom
- Fits130GB headroomest. 55.8 tok/s
- Fits386GB headroomest. 55.8 tok/s
| Config | Memory | Price | $/GB | Fit | Headroom | Est. speed |
|---|---|---|---|---|---|---|
| Mac mini, M6 | 16GB | $1,099 | $69 | Won't fit | −105GB | — |
| Mac mini, M6 | 24GB | $1,299 | $54 | Won't fit | −97GB | — |
| Mac mini, M6 | 32GB | $1,499 | $47 | Won't fit | −89GB | — |
| Mac mini, M5 Pro | 24GB | $1,699 | $71 | Won't fit | −97GB | — |
| Mac mini, M5 Pro | 48GB | $2,499 | $52 | Won't fit | −73GB | — |
| Mac Studio, M5 Max | 36GB | $2,499 | $69 | Won't fit | −85GB | — |
| Mac mini, M5 Pro | 64GB | $2,899 | $45 | Won't fit | −59GB | — |
| GeForce RTX 5090 | 32GB | $2,999.99+host | $94 | Won't fit | −87GB | — |
| Mac Studio, M5 Max | 48GB | $3,099 | $65 | Won't fit | −73GB | — |
| Mac Studio, M5 Max | 64GB | $3,499 | $55 | Won't fit | −59GB | — |
| Ascent GX10, GB10 Grace Blackwell | 128GB | $3,999.99 | $31 | Tight | 3GB | 12.7 tok/s |
| DGX Spark, GB10 Grace Blackwell | 128GB | $4,699 | $37 | Tight | 3GB | 12.7 tok/s |
| Mac Studio, M5 Max | 128GB | $5,099 | $40 | Tight | 2GB | 28.6 tok/s |
| Mac Studio, M5 Ultra | 96GB | $5,499 | $57 | Won't fit | −29GB | — |
| Mac Studio, M5 Ultra | 256GB | $9,499 | $37 | Fits | 130GB | 55.8 tok/s |
| Mac Studio, M5 Ultra | 512GB | — | — | Fits | 386GB | 55.8 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 = 235B × 0.489 B/param (3.91 bits) + 0.193 GB/1K × 8K K/V + 1.2 GB runtime
speed = bandwidth × 0.5 ÷ (22B active × 0.489 B/param) = 10.8 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 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.