aliteq.

Apple M5 Ultra: 512 GB, what it runs and what it costs

Specs checked · Aliteq Team · How we compute this

The short answer

The Apple M5 Ultra has 512 GB of VRAM and 1200 GB/s of memory bandwidth (2026). At Q4_K_M quality or better with an 8K context it runs 57 of the open models we track; the largest that fits with room to spare is MiniMax-M3.1-Flash (MoE). We don't see it for rent on the providers we track right now.

VRAM
512 GB
Bandwidth
1200 GB/s
Board power
—
Launched
2026

Source: Apple specification · bandwidth not yet confirmed by a second source, so we don't estimate speed from it

Buy it or rent it

Not for rent on the providers we track right now. See every card that is.

Which AI models run on the Apple M5 Ultra

Computed by our VRAM engine for an 8K context: the best quality each model fits at in 512 GB. Speeds are upper-bound estimates from memory bandwidth, not benchmarks.

ModelBest fitVRAM used~tok/s
DeepSeek V4.1 Flash (MoE)FP4 + FP8476 GB93% · tight—
GLM-5.3 (MoE)Q5_K_M500 GB98% · tight—
MiniMax-M3.1-Flash (MoE)NVFP4235 GB46%—
MiniMax-M3 (MoE)Q8_0424 GB83%—
Qwen3 235B-A22B (MoE)F16440 GB86%—
MiniMax-M2.5 (MoE)FP8217 GB42%—
Qwen3.8-Flash-Next (MoE)F16336 GB66%—
Qwen3.5 122B-A10B (MoE)F16234 GB46%—
NVIDIA Nemotron 3.5 Super VL 120B-A12B (MoE)F16233 GB45%—
NVIDIA Nemotron 3 Super 120B-A12B (MoE)F16231 GB45%—
Mistral Small 4 119B (MoE)FP8114 GB22%—
gpt-oss 120B (MoE)MXFP462 GB12%—

Too big for 512 GB (rent a bigger card instead): what it costs to run DeepSeek V4 Pro (MoE), what it costs to run Kimi K3 (MoE).

Check a model and context of your own in the VRAM calculator.

Our guides on the Apple M5 Ultra

Cards to compare it with

Every card by VRAM tier: all GPUs for gaming and local AI.

Apple M5 Ultra: common questions

How much VRAM does the Apple M5 Ultra have?
512 GB, with 1200 GB/s of memory bandwidth, per the maker's specification.
What is the biggest AI model the Apple M5 Ultra can run?
Of the open models we track, MiniMax-M3.1-Flash (MoE) is the largest that fits at Q4_K_M or better with an 8K context and room to spare (about 235 GB of 512 GB). DeepSeek V4.1 Flash (MoE) also squeezes in, but with almost no memory left for longer context. Bigger models only fit at lower-quality quants or need more memory.
Can I rent an Apple M5 Ultra in the cloud?
We don't see it on Vast.ai, Runpod or the other providers we track right now. The cloud GPU table lists every card that is for rent.
Is 512 GB enough?
Datacenter GPUs and unified-memory machines; mostly rented by the hour.