Mac Studio, M3 Ultra, 256GB, refurbished

1TB of storage, 819GB/s of memory bandwidth, and 248GB of unified memory left for a model once macOS has taken its share. It holds 18 of the 22 models we track.

The cheapest 256GB of unified memory Apple sells in any condition. Apple did not show an original price on this build, so we do not quote a saving against one.

Offers

checked 57 minutes ago

Apple Certified Refurbishedcheapest todayin stockchecked 57 minutes ago$7,729
Refurbished by Apple, with Apple's one-year limited warranty and 14-day returns. Apple says refurbished supply is usually very limited and often runs out, so this row can go dead faster than a new one.

No reseller undercuts Apple Certified Refurbished on this Config, so Apple Certified Refurbished direct is the price to beat.

Timing

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

Models it runs — 8K context

Llama-4-Behemoth—
Won't fit— headroom
Kimi-K2-Instruct—
Won't fit— headroom
DeepSeek-V3—
Won't fit— headroom
DeepSeek-R1—
Won't fit— headroom
Qwen3-Coder-480B-A35B-InstructQ3_K_M
Tight10GB headroomest. 23.9 tok/s
Llama-4-Maverick-17B-128EQ4_K_M
Tight4GB headroomest. 39.9 tok/s
GLM-4.6Q4_K_M
Fits28GB headroomest. 21.2 tok/s
Qwen3-235B-A22BQ6_K
Fits53GB headroomest. 22.7 tok/s
MiniMax-M2Q8_0
Tight1GB headroomest. 38.5 tok/s
Mistral-Large-2411Q8_0
Fits113GB headroomest. 4.1 tok/s
gpt-oss-120bBF16
Tight12GB headroomest. 40.1 tok/s
Llama-4-Scout-17B-16EBF16
Fits27GB headroomest. 12.0 tok/s
Llama-3.3-70BBF16
Fits104GB headroomest. 3.8 tok/s
Qwen3-32BBF16
Fits181GB headroomest. 8.3 tok/s
Qwen3-30B-A3BBF16
Fits185GB headroomest. 62.0 tok/s
Gemma-3-27BBF16
Fits191GB headroomest. 9.9 tok/s
Devstral-Small-2507BF16
Fits197GB headroomest. 11.1 tok/s
gpt-oss-20bBF16
Fits204GB headroomest. 56.9 tok/s
Phi-4BF16
Fits217GB headroomest. 19.0 tok/s
Qwen3-VL-8B-InstructBF16
Fits230GB headroomest. 33.3 tok/s
Gemma-3n-E4BBF16
Fits233GB headroomest. 66.5 tok/s
Qwen3-4BBF16
Fits238GB headroomest. 66.5 tok/s

Model	Best quant	Fit	Headroom	Est. speed
Llama-4-Behemoth	—	Won't fit	—	—
Kimi-K2-Instruct	—	Won't fit	—	—
DeepSeek-V3	—	Won't fit	—	—
DeepSeek-R1	—	Won't fit	—	—
Qwen3-Coder-480B-A35B-Instruct	Q3_K_M	Tight	10GB	23.9 tok/s
Llama-4-Maverick-17B-128E	Q4_K_M	Tight	4GB	39.9 tok/s
GLM-4.6	Q4_K_M	Fits	28GB	21.2 tok/s
Qwen3-235B-A22B	Q6_K	Fits	53GB	22.7 tok/s
MiniMax-M2	Q8_0	Tight	1GB	38.5 tok/s
Mistral-Large-2411	Q8_0	Fits	113GB	4.1 tok/s
gpt-oss-120b	BF16	Tight	12GB	40.1 tok/s
Llama-4-Scout-17B-16E	BF16	Fits	27GB	12.0 tok/s
Llama-3.3-70B	BF16	Fits	104GB	3.8 tok/s
Qwen3-32B	BF16	Fits	181GB	8.3 tok/s
Qwen3-30B-A3B	BF16	Fits	185GB	62.0 tok/s
Gemma-3-27B	BF16	Fits	191GB	9.9 tok/s
Devstral-Small-2507	BF16	Fits	197GB	11.1 tok/s
gpt-oss-20b	BF16	Fits	204GB	56.9 tok/s
Phi-4	BF16	Fits	217GB	19.0 tok/s
Qwen3-VL-8B-Instruct	BF16	Fits	230GB	33.3 tok/s
Gemma-3n-E4B	BF16	Fits	233GB	66.5 tok/s
Qwen3-4B	BF16	Fits	238GB	66.5 tok/s

Best quant is the highest-quality one that still loads with headroom. Every speed is an estimate; the formula is here.