Z.ai · 753B (40B active) · 專家混合（MoE）
Frontier open-weight coder — top SWE-bench, 1M context 看看你的 GPU 或 Mac 跑不跑得動 GLM-5.2——最低 420.8 GB，建議 701.3 GB。
2026-061024K context
專家混合（MoE）
專家總数： 256
啟用專家： 8
啟用参数： 40.0B
| 量化 | 位元 | VRAM | 品質 | 状態 |
|---|---|---|---|---|
| Q2_K | 2 | 241.6 GB | low | — |
| Q3_K_M | 3 | 338 GB | moderate | — |
| Q4_K_M | 4 | 386.2 GB | good | — |
| Q5_K_M | 5 | 482.6 GB | good | — |
| Q6_K | 6 | 579.1 GB | excellent | — |
| Q8_0 | 8 | 771.9 GB | excellent | — |
| F16 | 16 | 1543.3 GB | lossless | — |
Can I run GLM-5.2 locally?
- Can I run GLM-5.2 locally?
- GLM-5.2 needs about 420.8 GB of memory at a minimum and 701.3 GB recommended. Open this page to grade it against your GPU or Mac, then run it with runai, Ollama or LM Studio.
- How much VRAM does GLM-5.2 need?
- At Q4_K_M, GLM-5.2 uses about 386.2 GB of VRAM. Higher quants need more memory; lower quants fit tighter cards with a quality tradeoff.