How much memory does gpt-oss-20b at Q6_K need?

19GB. That is 17GB of weights at 6.56 bits per weight, 0.4GB of K/V cache at 8K context, and 1.2GB of runtime overhead.

Which machine is cheapest for gpt-oss-20b at Q6_K?

Mac mini, M6, 32GB, $1,499 at Apple. It fits with 10GB of headroom and decodes at an est. 28.8 tok/s.

Cheapest box to run gpt-oss-20b at Q6_K

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.

Fit finderprices checked 10 hours ago

Quant

Context

gpt-oss-20b at Q6_K needs 19GB.

Cheapest Config that fits: Mac mini, M6, 32GB. Fits, 10GB headroom. Est. 28.8 tok/s.

Best offer today: $1,499 at Apple — in stock, checked 10 hours ago.

Collecting price history — 2 of 90 days. No timing verdict yet.

Offers — Mac mini, M6, 32GB, 512GB

Applelist pricein stockchecked 10 hours ago$1,499
Expercomout of stockchecked 10 hours ago$1,499
B&H Photoout of stockchecked 10 hours ago$1,599

Mac mini, M6, 16GB$1,099
Won't fit−6GB headroom
Mac mini, M6, 24GB$1,299
Tight2GB headroomest. 28.8 tok/s
Mac mini, M6, 32GB$1,499
Fits10GB headroomest. 28.8 tok/s
Mac mini, M5 Pro, 24GB$1,699
Tight2GB headroomest. 52.0 tok/s
Mac mini, M5 Pro, 48GB$2,499
Fits25GB headroomest. 52.0 tok/s
Mac Studio, M5 Max, 36GB$2,499
Fits14GB headroomest. 104 tok/s
Mac mini, M5 Pro, 64GB$2,899
Fits40GB headroomest. 52.0 tok/s
GeForce RTX 5090, 32GB$2,999.99+host
Fits12GB headroomest. 304 tok/s
Mac Studio, M5 Max, 48GB$3,099
Fits25GB headroomest. 104 tok/s
Mac Studio, M5 Max, 64GB$3,499
Fits40GB headroomest. 104 tok/s
Ascent GX10, GB10 Grace Blackwell, 128GB$3,999.99
Fits102GB headroomest. 46.2 tok/s
DGX Spark, GB10 Grace Blackwell, 128GB$4,699
Fits102GB headroomest. 46.2 tok/s
Mac Studio, M5 Max, 128GB$5,099
Fits101GB headroomest. 104 tok/s
Mac Studio, M5 Ultra, 96GB$5,499
Fits70GB headroomest. 203 tok/s
Mac Studio, M5 Ultra, 256GB$9,499
Fits229GB headroomest. 203 tok/s
Mac Studio, M5 Ultra, 512GB—
Fits485GB headroomest. 203 tok/s

Config	Memory	Price	$/GB	Fit	Headroom	Est. speed
Mac mini, M6	16GB	$1,099	$69	Won't fit	−6GB	—
Mac mini, M6	24GB	$1,299	$54	Tight	2GB	28.8 tok/s
Mac mini, M6	32GB	$1,499	$47	Fits	10GB	28.8 tok/s
Mac mini, M5 Pro	24GB	$1,699	$71	Tight	2GB	52.0 tok/s
Mac mini, M5 Pro	48GB	$2,499	$52	Fits	25GB	52.0 tok/s
Mac Studio, M5 Max	36GB	$2,499	$69	Fits	14GB	104 tok/s
Mac mini, M5 Pro	64GB	$2,899	$45	Fits	40GB	52.0 tok/s
GeForce RTX 5090	32GB	$2,999.99+host	$94	Fits	12GB	304 tok/s
Mac Studio, M5 Max	48GB	$3,099	$65	Fits	25GB	104 tok/s
Mac Studio, M5 Max	64GB	$3,499	$55	Fits	40GB	104 tok/s
Ascent GX10, GB10 Grace Blackwell	128GB	$3,999.99	$31	Fits	102GB	46.2 tok/s
DGX Spark, GB10 Grace Blackwell	128GB	$4,699	$37	Fits	102GB	46.2 tok/s
Mac Studio, M5 Max	128GB	$5,099	$40	Fits	101GB	104 tok/s
Mac Studio, M5 Ultra	96GB	$5,499	$57	Fits	70GB	203 tok/s
Mac Studio, M5 Ultra	256GB	$9,499	$37	Fits	229GB	203 tok/s
Mac Studio, M5 Ultra	512GB	—	—	Fits	485GB	203 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 × 0.820 B/param (6.56 bits) + 0.049 GB/1K × 8K K/V + 1.2 GB runtime
speed = bandwidth × 0.5 ÷ (3.6B active × 0.820 B/param) = 3.0 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.

What else Mac mini M6 32GB runs

Qwen3-32BQ5_K_M, est. 4.9 tok/s
Gemma-3-27BQ6_K, est. 5.0 tok/s
gpt-oss-20bQ8_0, est. 22.2 tok/s

Best quant that still fits at 8K context, largest model first. Full list on the Config page.

Bandwidth 170GB/s, usable memory 29GB of 32GB.