Mac Studio, M4 Max, 128GB, refurbished vs Mac Studio, M3 Ultra, 256GB, refurbished
Both hold gpt-oss-120b at Q6_K. Mac Studio, M3 Ultra, 256GB, refurbished decodes an est. 97.9 tok/s against 65.3, and costs $3,320 more — $101.72 per extra token per second. Worth it if you read the output as it arrives; not if you queue jobs.
- Fit
- Fits
- Fits
- Headroom
- 22GB
- 150GB
- Est. speed
- 65.3 tok/s
- 97.9 tok/s
- Cheapest today
- $4,409 at Apple Certified Refurbished
- $7,729 at Apple Certified Refurbished
- Per GB of memory
- $34
- $30
- Memory
- 128GB unified memory
- 256GB unified memory
- Usable for a model
- 120GB
- 248GB
- Memory bandwidth
- 546GB/s
- 819GB/s
- Storage
- 1TB
- 1TB
- Lowest recorded
- $4,409 in 3d
- $7,729 in 3d
Fit, headroom and speed are computed from each vendor’s published bandwidth and the model’s own size at 8K context. Prices are the cheapest listing we could read on each seller’s page. The arithmetic is here.
Where they actually differ
- Qwen3-235B-A22BWon't fitFits
- DeepSeek-V3Won't fitWon't fit
- Llama-3.3-70BFitsFits
- gpt-oss-120bFitsFits
- Mistral-Large-2411FitsFits
- Llama-4-Scout-17B-16EFitsFits
- Qwen3-32BFitsFits
- Gemma-3-27BFitsFits
- gpt-oss-20bFitsFits
- Llama-4-Maverick-17B-128EWon't fitWon't fit
- Llama-4-BehemothWon't fitWon't fit
- Qwen3-Coder-480B-A35B-InstructWon't fitWon't fit
- Kimi-K2-InstructWon't fitWon't fit
- GLM-4.6Won't fitWon't fit
- DeepSeek-R1Won't fitWon't fit
- MiniMax-M2Won't fitFits
- Qwen3-30B-A3BFitsFits
- Qwen3-4BFitsFits
- Devstral-Small-2507FitsFits
- Phi-4FitsFits
- Gemma-3n-E4BFitsFits
- Qwen3-VL-8B-InstructFitsFits
Every model we track at Q6_K. The rows where the two boxes agree are dimmed, because they are not what you are choosing between.