Cheapest box to run gpt-oss-20b at Q5_K_M
OpenAI’s gpt-oss-20b is 21B parameters, 3.6B 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.
gpt-oss-20b at Q5_K_M needs 16GB.
Cheapest Config that fits: Mac mini, M6, 24GB. Fits, 5GB headroom. Est. 33.3 tok/s.
Best offer today: $1,299 at Apple — in stock, checked 10 hours ago.
Collecting price history — 2 of 90 days. No timing verdict yet.
Offers — Mac mini, M6, 24GB, 512GB
- Applelist pricein stockchecked 10 hours ago$1,299
- Mac mini, M6, 16GB$1,099Won't fit−3GB headroom
- Mac mini, M6, 24GB$1,299Fits5GB headroomest. 33.3 tok/s
- Mac mini, M6, 32GB$1,499Fits13GB headroomest. 33.3 tok/s
- Mac mini, M5 Pro, 24GB$1,699Fits5GB headroomest. 60.2 tok/s
- Mac mini, M5 Pro, 48GB$2,499Fits28GB headroomest. 60.2 tok/s
- Mac Studio, M5 Max, 36GB$2,499Fits17GB headroomest. 120 tok/s
- Mac mini, M5 Pro, 64GB$2,899Fits42GB headroomest. 60.2 tok/s
- GeForce RTX 5090, 32GB$2,999.99+hostFits14GB headroomest. 351 tok/s
- Mac Studio, M5 Max, 48GB$3,099Fits28GB headroomest. 120 tok/s
- Mac Studio, M5 Max, 64GB$3,499Fits42GB headroomest. 120 tok/s
- Fits104GB headroomest. 53.5 tok/s
- Fits104GB headroomest. 53.5 tok/s
- Fits104GB headroomest. 120 tok/s
- Fits72GB headroomest. 235 tok/s
- Fits232GB headroomest. 235 tok/s
- Fits488GB headroomest. 235 tok/s
| Config | Memory | Price | $/GB | Fit | Headroom | Est. speed |
|---|---|---|---|---|---|---|
| Mac mini, M6 | 16GB | $1,099 | $69 | Won't fit | −3GB | — |
| Mac mini, M6 | 24GB | $1,299 | $54 | Fits | 5GB | 33.3 tok/s |
| Mac mini, M6 | 32GB | $1,499 | $47 | Fits | 13GB | 33.3 tok/s |
| Mac mini, M5 Pro | 24GB | $1,699 | $71 | Fits | 5GB | 60.2 tok/s |
| Mac mini, M5 Pro | 48GB | $2,499 | $52 | Fits | 28GB | 60.2 tok/s |
| Mac Studio, M5 Max | 36GB | $2,499 | $69 | Fits | 17GB | 120 tok/s |
| Mac mini, M5 Pro | 64GB | $2,899 | $45 | Fits | 42GB | 60.2 tok/s |
| GeForce RTX 5090 | 32GB | $2,999.99+host | $94 | Fits | 14GB | 351 tok/s |
| Mac Studio, M5 Max | 48GB | $3,099 | $65 | Fits | 28GB | 120 tok/s |
| Mac Studio, M5 Max | 64GB | $3,499 | $55 | Fits | 42GB | 120 tok/s |
| Ascent GX10, GB10 Grace Blackwell | 128GB | $3,999.99 | $31 | Fits | 104GB | 53.5 tok/s |
| DGX Spark, GB10 Grace Blackwell | 128GB | $4,699 | $37 | Fits | 104GB | 53.5 tok/s |
| Mac Studio, M5 Max | 128GB | $5,099 | $40 | Fits | 104GB | 120 tok/s |
| Mac Studio, M5 Ultra | 96GB | $5,499 | $57 | Fits | 72GB | 235 tok/s |
| Mac Studio, M5 Ultra | 256GB | $9,499 | $37 | Fits | 232GB | 235 tok/s |
| Mac Studio, M5 Ultra | 512GB | — | — | Fits | 488GB | 235 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 = 21B × 0.709 B/param (5.67 bits) + 0.049 GB/1K × 8K K/V + 1.2 GB runtime
speed = bandwidth × 0.5 ÷ (3.6B active × 0.709 B/param) = 2.6 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 M6 24GB runs
- Qwen3-32BQ3_K_M, est. 7.1 tok/s
- Gemma-3-27BQ4_K_M, est. 6.8 tok/s
- gpt-oss-20bQ6_K, est. 28.8 tok/s
Best quant that still fits at 8K context, largest model first. Full list on the Config page.
Bandwidth 170GB/s, usable memory 21GB of 24GB.