The real cost to run GLM 5.2 504B

0xSero · 290B parameters · 1M context · commercial OK

GLM 5.2 504B — 290B params (0xSero). Auto-indexed from the Hugging Face Hub (11,773 downloads). Parameter count is exact; download size and quantizations are estimates.

What it costs to run GLM 5.2 504B — $0 markup

On your own machine — $0. Runs free locally if you have about 227 GB of VRAM at Q4_K_M (point Spanvero at LM Studio, Ollama, or llama.cpp).
On your own rented GPU — from $2.94/hr. 6× NVIDIA RTX A6000 48GB at the direct RunPod price (≈ $3.67 a session), $0 markup.
Via your own API key — $2.42/1M tokens (blended input + output). No GPU to manage (rough estimate for this size).

Key facts

Parameters	290B
Context window	1M tokens
Recommended quant	Q4_K_M
VRAM to run	~227 GB (at Q4_K_M, 16.4K context)
Download size	~174 GB
License	Commercial use OK

Open the free Spanvero advisor → for the live, interactive math for your exact workload and hardware.

Related models

Hy3 preview — 298.8B, tencent
MiMo V2 Flash — 309.8B, XiaomiMiMo
MiniMax M3 NVFP4 — 246.6B, nvidia
NVIDIA Nemotron 3 Ultra 550B A55B NVFP4 — 335B, nvidia
Qwen3 235B A22B Instruct 2507 — 235.1B, Qwen

Browse: All models · Compare