GLM 4.5 needs about 220.1 GB of memory at Q4_K_M (a 216.5 GB file) with an 8K context window, which is more than a 48 GB GPU holds. This repository has 27 quants, from 84.3 GB (TQ1_0) to 716.8 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 12.5 GB.
GLM‑4.5, a 355‑billion‑parameter foundation model for intelligent agents
GLM‑4.5 is a 355‑billion‑parameter foundation model designed for intelligent agents. It supports hybrid reasoning with thinking and non‑thinking modes. The model is released under the MIT license and can be used commercially.
Good at
- Hybrid reasoning and tool usage
- Large context handling
- Agent‑centric tasks
Limits
- No vision support
- Requires large VRAM for full precision
- License is MIT (open source)
Best for
- Developing intelligent agents
- Complex reasoning workflows
- Tool‑integrated applications
Who it’s for. Developers building intelligent agent systems who need a large‑scale language model with hybrid reasoning capabilities. The model is suitable for commercial use under the MIT license.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| TQ1_0 | 84.3 GB | 88.0 GB | Too big | Get file → | |
| IQ1_S | 97.0 GB (2 parts) | 100.7 GB | Too big | Get file → | |
| IQ1_M | 107.7 GB (3 parts) | 111.3 GB | Too big | Get file → | |
| IQ2_XXS | 115.9 GB (3 parts) | 119.5 GB | Too big | Get file → | |
| IQ2_M | 122.4 GB (3 parts) | 126.1 GB | Too big | Get file → | |
| Q2_K | 131.2 GB (3 parts) | 134.8 GB | Too big | Get file → | |
| Q2_K_L | 131.3 GB (3 parts) | 135.0 GB | Too big | Get file → | |
| Q2_K_XL | 135.2 GB (3 parts) | 138.8 GB | Too big | Get file → | |
| IQ3_XXS | 145.2 GB (3 parts) | 148.8 GB | Too big | Get file → | |
| Q3_K_S | 155.0 GB (4 parts) | 158.6 GB | Too big | Get file → | |
| Q3_K_XL | 158.7 GB (4 parts) | 162.4 GB | Too big | Get file → | |
| Q3_K_M | 171.3 GB (4 parts) | 174.9 GB | Too big | Get file → | |
| IQ4_XS | 191.6 GB (4 parts) | 195.2 GB | Too big | Get file → | |
| IQ4_NL | 202.5 GB (5 parts) | 206.1 GB | Too big | Get file → | |
| Q4_0 | 203.0 GB (5 parts) | 206.7 GB | Too big | Get file → | |
| Q4_K_S | 203.7 GB (5 parts) | 207.3 GB | Too big | Get file → | |
| Q4_K_XL | 204.4 GB (5 parts) | 208.0 GB | Too big | Get file → | |
| Q4_K_M | 216.5 GB (5 parts) | 220.1 GB | Too big | Get file → | |
| Q4_1 | 224.6 GB (5 parts) | 228.2 GB | Too big | Get file → | |
| Q5_K_S | 247.0 GB (5 parts) | 250.6 GB | Too big | Get file → | |
| Q5_K_XL | 253.1 GB (6 parts) | 256.7 GB | Too big | Get file → | |
| Q5_K_M | 254.2 GB (6 parts) | 257.9 GB | Too big | Get file → | |
| Q6_K | 294.4 GB (6 parts) | 298.0 GB | Too big | Get file → | |
| Q6_K_XL | 300.2 GB (7 parts) | 303.9 GB | Too big | Get file → | |
| Q8_0 | 381.0 GB (8 parts) | 384.6 GB | Too big | Get file → | |
| Q8_K_XL | 391.5 GB (8 parts) | 395.1 GB | Too big | Get file → | |
| BF16 | 716.8 GB (15 parts) | 720.5 GB | Too big | Get file → |
For 8 GB VRAM, use Q4_K_S or Q4_K_XL; they estimate 207 GB memory at 8 k context. For 16 GB VRAM, Q5_K_S or Q5_K_XL are suitable, estimating 250 GB memory. For larger budgets, Q6_K or Q8_0 provide higher fidelity.
What GPU do you need for GLM 4.5?
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 4.5 quant should you choose?
Q4_K_M is the usual starting point for GLM 4.5: it balances file size and quality well. Here it is a 216.5 GB file.
The Q8_0 file is 381.0 GB, about 1.8 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 TQ1_0 at 84.3 GB, with the quality trade-offs that come with fewer bits. Lower bit counts mean smaller files and more quality loss.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about GLM 4.5
How much VRAM does GLM 4.5 need?
About 220.1 GB at Q4_K_M with an 8K context window. That is the 216.5 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 4.5 on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 88.0 GB, so you would need a larger GPU or CPU offloading.
What is the file size of GLM 4.5 GGUF?
It depends on the quant: TQ1_0 is 84.3 GB, IQ1_S is 97.0 GB, IQ1_M is 107.7 GB, IQ2_XXS is 115.9 GB, IQ2_M is 122.4 GB, Q2_K is 131.2 GB and 21 more.
What context length does GLM 4.5 support?
The GGUF header lists a trained context length of 131,072 tokens (128K). The KV cache grows with the context you use: about 3.1 GB at 8K and 12.5 GB at 32K.
Where can I download GLM 4.5 GGUF?
The files are on Hugging Face in the unsloth/GLM-4.5-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 4.5 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‑4.5 on 8 GB VRAM?
Only with low‑precision quants such as Q4_K_S or Q4_K_XL, which estimate 207 GB memory at 8 k context.
Does GLM‑4.5 support vision tasks?
No, the model has no vision capabilities.