GLM 5.2 needs about 467.7 GB of memory at Q4_K_M (a 465.8 GB file) with an 8K context window, which is more than a 48 GB GPU holds. This repository has 22 quants, from 216.7 GB (IQ1_S) to 1508.0 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 5.6 GB.
GLM‑5.2, a 750‑billion‑parameter LLM with a 1 M‑token context
GLM‑5.2 is a 750 billion‑parameter model built on the glm‑dsa architecture. It supports a solid 1 M‑token context and offers advanced coding and speculative decoding features. The model is released under an MIT license and is available as a set of GGUF files.
Good at
- Long‑horizon reasoning with 1 M context
- Flexible coding effort levels
- Speculative decoding for faster inference
Limits
- Requires large VRAM for full‑precision; high memory for 8 k context
- No vision or multimodal support
- MIT license limits no commercial restrictions
Best for
- Research on long‑form text generation
- Complex coding tasks with adjustable latency
- Large‑scale inference on high‑end GPUs
Who it’s for. Researchers and developers needing extensive context for reasoning or coding tasks. The model is open source and suitable for academic or non‑commercial use.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| IQ1_S | 216.7 GB (6 parts) | 218.6 GB | Too big | Get file → | |
| IQ1_M | 228.5 GB (6 parts) | 230.4 GB | Too big | Get file → | |
| IQ2_XXS | 238.5 GB (6 parts) | 240.4 GB | Too big | Get file → | |
| IQ2_M | 238.6 GB (6 parts) | 240.5 GB | Too big | Get file → | |
| Q2_K_XL | 253.9 GB (7 parts) | 255.8 GB | Too big | Get file → | |
| IQ3_XXS | 281.7 GB (7 parts) | 283.6 GB | Too big | Get file → | |
| IQ3_S | 308.6 GB (8 parts) | 310.5 GB | Too big | Get file → | |
| Q3_K_M | 342.7 GB (9 parts) | 344.6 GB | Too big | Get file → | |
| Q3_K_XL | 343.0 GB (9 parts) | 344.9 GB | Too big | Get file → | |
| IQ4_XS | 365.3 GB (9 parts) | 367.2 GB | Too big | Get file → | |
| IQ4_NL | 372.7 GB (9 parts) | 374.6 GB | Too big | Get file → | |
| Q4_K_S | 436.4 GB (10 parts) | 438.3 GB | Too big | Get file → | |
| Q4_K_M | 465.8 GB (11 parts) | 467.7 GB | Too big | Get file → | |
| Q4_K_XL | 467.3 GB (11 parts) | 469.2 GB | Too big | Get file → | |
| Q5_K_S | 527.3 GB (12 parts) | 529.2 GB | Too big | Get file → | |
| Q5_K_M | 560.8 GB (13 parts) | 562.7 GB | Too big | Get file → | |
| Q5_K_XL | 562.5 GB (13 parts) | 564.4 GB | Too big | Get file → | |
| Q6_K | 625.9 GB (14 parts) | 627.8 GB | Too big | Get file → | |
| Q6_K_XL | 684.4 GB (16 parts) | 686.3 GB | Too big | Get file → | |
| Q8_0 | 801.4 GB (17 parts) | 803.3 GB | Too big | Get file → | |
| Q8_K_XL | 819.7 GB (18 parts) | 821.6 GB | Too big | Get file → | |
| BF16 | 1508.0 GB (33 parts) | 1509.9 GB | Too big | Get file → |
For 8 GB VRAM, use IQ1_S (216 GB file, ~219 GB memory). For 16 GB VRAM, Q4_K_S (436 GB file, ~438 GB memory). For 24 GB VRAM, Q5_K_M (560 GB file, ~563 GB memory). Memory estimates are approximate.
What GPU do you need for GLM 5.2?
These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.
- 4 GB VRAM: no quant fits at an 8K context.
- 6 GB VRAM: no quant fits at an 8K context.
- 8 GB VRAM: no quant fits at an 8K context.
- 12 GB VRAM: no quant fits at an 8K context.
- 16 GB VRAM: no quant fits at an 8K context.
- 24 GB VRAM: no quant fits at an 8K context.
- 32 GB VRAM: no quant fits at an 8K context.
- 48 GB VRAM: no quant fits at an 8K context.
Which GLM 5.2 quant should you choose?
Q4_K_M is the usual starting point for GLM 5.2: it balances file size and quality well. Here it is a 465.8 GB file.
The Q8_0 file is 801.4 GB, about 1.7 times the size of Q4_K_M. It keeps the weights closer to the original, which only matters if you have the memory to spare.
If memory is tight, the smallest file is IQ1_S at 216.7 GB, with the quality trade-offs that come with fewer bits. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about GLM 5.2
How much VRAM does GLM 5.2 need?
About 467.7 GB at Q4_K_M with an 8K context window. That is the 465.8 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.
Can I run GLM 5.2 on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 218.6 GB, so you would need a larger GPU or CPU offloading.
What is the file size of GLM 5.2 GGUF?
It depends on the quant: IQ1_S is 216.7 GB, IQ1_M is 228.5 GB, IQ2_XXS is 238.5 GB, IQ2_M is 238.6 GB, Q2_K_XL is 253.9 GB, IQ3_XXS is 281.7 GB and 16 more.
What context length does GLM 5.2 support?
The GGUF header lists a trained context length of 1,048,576 tokens (1024K). The KV cache grows with the context you use: about 1.4 GB at 8K and 5.6 GB at 32K.
Where can I download GLM 5.2 GGUF?
The files are on Hugging Face in the unsloth/GLM-5.2-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does GLM 5.2 use?
Hugging Face lists the license as mit. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.
Can I run GLM‑5.2 on 8 GB VRAM?
Only the lowest‑precision quantizations such as IQ1_S are feasible; higher quants require more VRAM.
Does GLM‑5.2 support vision tasks?
No, the model is purely text‑based and has no vision or multimodal capabilities.