CHEAPEST HARDWARE FOR 70B-CLASS MODELS

answer: Minisforum MS-S1 MAX at $2,919.90 is the cheapest single device here that fits Llama 3.3 70B Q4_K_M (needs ≈49 GB). No single consumer GPU in the catalog has enough VRAM (the RTX 5090 has 32GB).
gpu_route: two or more GPUs, or a lower quant (Q3/Q2) with quality loss — not rated here
buy: BUY NOW → Minisforum

test_model: Llama 3.3 70B Instruct, Q4_K_M GGUF, 42.5 GB file, needs ≈49.5 GB with 8k context
rank_by: lowest current official price among devices where it fits
prices: official maker list / MSRP only, latest recorded point, linked in SOURCES
fit: weights file size (Hugging Face) + KV cache at 8,192 tokens (model config.json) + 10% overhead, compared with usable memory: VRAM − 1 GB, or 75% of unified / system memory (details)
speed: only published llama.cpp runs; we never estimate tok/s
not_considered: street prices, used hardware, multi-GPU builds, power and noise