RunsOnMy ModelsGPUs & MacsBy VRAMBest GPUCan I run…?OllamaStart hereAPI pricesVRAM calculatorNew

2x RTX 3090 vs Mac M3 Ultra 256GB for local LLMs

Pick the Mac M3 Ultra 256GB if you want to run bigger models (it runs 49 of the top 60 vs 43); pick the 2x RTX 3090 if the models you use already fit in it and you want more tokens per second.

Specs that matter for LLMs

2x RTX 3090Mac M3 Ultra 256GB
Memory (VRAM)2×24 GB256 GB unified
Usable for a model (our estimate)45.6 GB192.0 GB
Memory bandwidth936 GB/s819 GB/s
Launch MSRP$2,998—
Launch $ per GB of memory$62—
Top 60 models that fit (4-bit or better, 8K ctx)4349
…of those at 8-bit or better4345
Largest top-60 model that fitsQwen3.5 35B A3BHy3
Offload headroom: more top-60 models via llama.cpp CPU offload (≥3 tok/s est., up to 128 GB RAM)not modelled for multi-GPU setupsn/a: one shared memory pool

Top 60 = the 60 most-downloaded open models we track. Launch MSRP from vendor announcements; Macs and workstation parts have no single launch price, so none is shown. How we calculate.

Estimated speed on popular models

Best quant that fits at 8K context, with a tokens/sec ceiling from memory bandwidth (real speed is typically 50–80% of it). "Offload" means part of the model sits in system RAM. Estimates, not benchmarks.

ModelNeeds at 4-bit2x RTX 3090Mac M3 Ultra 256GB
Qwen3 8B6.5 GBF16 · ~57 tok/sF16 · ~50 tok/s
Llama 3.1 8B Instruct6.2 GBF16 · ~58 tok/sF16 · ~51 tok/s
Qwen3 14B10.3 GBF16 · ~32 tok/sF16 · ~28 tok/s
gpt-oss 20B11.8 GBF16 · ~338 tok/sF16 · ~296 tok/s
Gemma 3 27B IT17.3 GBQ8_0 · ~33 tok/sF16 · ~15 tok/s
Qwen3 32B21.1 GBQ8_0 · ~27 tok/sF16 · ~12 tok/s
Qwen3 Coder 30B A3B Instruct18.8 GBQ8_0 · ~293 tok/sF16 · ~137 tok/s
Llama 3.3 70B Instruct42.7 GBQ3_K_M only (below 4-bit)F16 · ~6 tok/s
gpt-oss 120B59.6 GBNoF16 · ~255 tok/s

Models only one of them runs

Only the 2x RTX 3090

None.

Only the Mac M3 Ultra 256GB

More comparisons

All GPU comparisons

See every model on the 2x RTX 3090 or the Mac M3 Ultra 256GB, or try your own settings in the VRAM calculator.