GLM 4.7 Flash needs about 19.8 GB of memory at Q4_K_M (a 18.5 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 28 quants, from 0.1 GB to 59.9 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 3.4 GB.
GLM 4.7 Flash – 30B‑A3B MoE model for efficient local inference
GLM 4.7 Flash is a 30B‑A3B mixture‑of‑experts model from the GLM family. It supports inference with llama.cpp, vLLM, and SGLang. The model is released under the MIT license and offers a balance of performance and memory efficiency.
Good at
- High‑capacity context handling
- MoE‑based specialization
- Efficient inference with low‑bit quantization
Limits
- Requires 8‑k context for memory estimates
- No vision support
- Large memory footprint for full precision
Best for
- Local LLM deployment on GPUs
- Research prototyping with MoE models
- Fine‑tuning lightweight variants
Who it’s for. Engineers and researchers who need a 30B‑class model that can run locally with llama.cpp or similar frameworks. The model is suitable for those who value a balance between performance and memory usage.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| — | 0.1 GB | 1.4 GB | Fits | Get file → | |
| IQ2_XXS | 7.6 GB | 9.0 GB | Too big | Get file → | |
| IQ2_XS | 8.7 GB | 10.0 GB | Too big | Get file → | |
| IQ2_S | 8.8 GB | 10.1 GB | Too big | Get file → | |
| IQ2_M | 9.9 GB | 11.2 GB | Too big | Get file → | |
| Q2_K | 11.0 GB | 12.4 GB | Too big | Get file → | |
| Q2_K_L | 11.4 GB | 12.7 GB | Too big | Get file → | |
| IQ3_XXS | 12.2 GB | 13.6 GB | Too big | Get file → | |
| IQ3_XS | 12.7 GB | 14.0 GB | Too big | Get file → | |
| Q3_K_S | 13.5 GB | 14.8 GB | Too big | Get file → | |
| IQ3_M | 14.0 GB | 15.3 GB | Too big | Get file → | |
| Q3_K_M | 14.1 GB | 15.4 GB | Too big | Get file → | |
| Q3_K_L | 14.5 GB | 15.8 GB | Too big | Get file → | |
| Q3_K_XL | 14.8 GB | 16.1 GB | Too big | Get file → | |
| IQ4_XS | 16.3 GB | 17.6 GB | Too big | Get file → | |
| IQ4_NL | 17.1 GB | 18.5 GB | Too big | Get file → | |
| Q4_0 | 17.4 GB | 18.7 GB | Too big | Get file → | |
| Q4_K_S | 17.8 GB | 19.1 GB | Too big | Get file → | |
| Q4_K_M | 18.5 GB | 19.8 GB | Too big | Get file → | |
| Q4_K_L | 18.7 GB | 20.0 GB | Too big | Get file → | |
| Q4_1 | 19.0 GB | 20.3 GB | Too big | Get file → | |
| Q5_K_S | 20.8 GB | 22.2 GB | Too big | Get file → | |
| Q5_K_M | 21.6 GB | 22.9 GB | Too big | Get file → | |
| Q5_K_L | 21.8 GB | 23.1 GB | Too big | Get file → | |
| Q6_K | 24.8 GB | 26.2 GB | Too big | Get file → | |
| Q6_K_L | 25.0 GB | 26.3 GB | Too big | Get file → | |
| Q8_0 | 31.8 GB | 33.2 GB | Too big | Get file → | |
| BF16 | 59.9 GB (2 parts) | 61.2 GB | Too big | Get file → |
For 8 GB VRAM, use IQ2_XXS (7.62 GB) or IQ2_XS (8.67 GB). For 16 GB VRAM, Q4_K_M (18.47 GB) or Q4_K_L (18.71 GB) provide a good trade‑off. For 32 GB VRAM, Q6_K (24.83 GB) or Q6_K_L (24.98 GB) are recommended. Memory estimates are approximate.
What GPU do you need for GLM 4.7 Flash?
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: the best quant that fits is the main file (0.1 GB file, about 1.4 GB in use).
- 6 GB VRAM: the best quant that fits is the main file (0.1 GB file, about 1.4 GB in use).
- 8 GB VRAM: the best quant that fits is the main file (0.1 GB file, about 1.4 GB in use).
- 12 GB VRAM: the best quant that fits is IQ2_M (9.9 GB file, about 11.2 GB in use).
- 16 GB VRAM: the best quant that fits is Q3_K_L (14.5 GB file, about 15.8 GB in use).
- 24 GB VRAM: the best quant that fits is Q5_K_L (21.8 GB file, about 23.1 GB in use).
- 32 GB VRAM: the best quant that fits is Q6_K_L (25.0 GB file, about 26.3 GB in use).
- 48 GB VRAM: the best quant that fits is Q8_0 (31.8 GB file, about 33.2 GB in use).
Which GLM 4.7 Flash quant should you choose?
Q4_K_M is the usual starting point for GLM 4.7 Flash: it balances file size and quality well. Here it is a 18.5 GB file.
The Q8_0 file is 31.8 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 the first one in the table at 0.1 GB, with the quality trade-offs that come with fewer bits.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about GLM 4.7 Flash
How much VRAM does GLM 4.7 Flash need?
About 19.8 GB at Q4_K_M with an 8K context window. That is the 18.5 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 4.7 Flash on an 8 GB GPU?
Yes, with the smaller files: the 0.1 GB file needs about 1.4 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of GLM 4.7 Flash GGUF?
It depends on the quant: file is 0.1 GB, IQ2_XXS is 7.6 GB, IQ2_XS is 8.7 GB, IQ2_S is 8.8 GB, IQ2_M is 9.9 GB, Q2_K is 11.0 GB and 22 more.
What context length does GLM 4.7 Flash support?
The GGUF header lists a trained context length of 202,752 tokens (198K). The KV cache grows with the context you use: about 0.8 GB at 8K and 3.4 GB at 32K.
Where can I download GLM 4.7 Flash GGUF?
The files are on Hugging Face in the bartowski/zai-org_GLM-4.7-Flash-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 4.7 Flash 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 4.7 Flash on 8 GB VRAM?
Yes, using IQ2_XXS or IQ2_XS quantizations, which require about 9–10 GB of VRAM at 8k context.
Does the model support vision tasks?
No, GLM 4.7 Flash does not have vision capabilities.
What license applies to this model?
The model is released under the MIT license.