GLM-4.7 GGUF size and VRAM requirements

General Purpose LLM License: mit ⬇ 11,811 ❤ 224
Parameters358.34B
Context202,752

GLM-4.7-GGUF is a very large language model focused on general purpose & instruction. With approximately 358.34 billion parameters, it balances local hardware requirements with reasoning and generation fidelity. It is built on the glm4moe architecture with a native context window of up to 202,752 tokens. The model is primarily recommended for general knowledge, document summarization, email drafting, language translation, and multi-turn conversational chat. Released under the mit license, it can be executed fully offline without sending data to external APIs.

GLM-4.7-GGUF is a Mixture-of-Experts model with 160 experts, of which 8 are active on each token. Routing only 8 of 160 experts makes it noticeably faster than a dense model of the same size — but every expert still has to be held in memory, so the numbers below are set by the full parameter count, not the active one. It is built from 92 transformer layers, a hidden size of 5,120, 96 attention heads. It uses grouped-query attention (96 query heads sharing 8 key/value heads), which already trims KV-cache memory compared with full multi-head attention.

To run unsloth/GLM-4.7-GGUF locally at a 4,096-token context, its quantized versions need between 80.93 GB (TQ1_0, lowest quality) and 669.84 GB (BF16, highest quality) of memory, weights plus KV cache and a system margin included. Context length drives that memory directly: at 4,096 tokens the KV cache for GLM-4.7-GGUF is about 1.44 GB, rising to roughly 71.16 GB at its full 202,752-token context. Shorter prompts free up memory for a higher-quality quantization.

Available GGUF quantizations for unsloth/GLM-4.7-GGUF include TQ1_0, IQ1_S, IQ1_M, IQ2_XXS, IQ2_M, Q2_K, Q2_K_L, Q2_K_XL, IQ3_XXS, Q3_K_S, Q3_K_XL, Q3_K_M, IQ4_XS, IQ4_NL, Q4_0, Q4_K_S, Q4_K_XL, Q4_K_M, Q4_1, Q5_K_S, Q5_K_XL, Q5_K_M, Q6_K, Q6_K_XL, Q8_0, Q8_K_XL, BF16. The model supports a native context length of up to 202,752 tokens; a longer context grows the KV cache and the memory needed.

→ Guide: How much VRAM do you need?

General Purpose LLM

Recommended Use Cases & Local Setup

General Purpose & Instruction

Ideal for: General knowledge, document summarization, email drafting, language translation, and multi-turn conversational chat.

Ollama

Simple one-line CLI installation running silently in the background

LM Studio

Polished desktop client with one-click model downloads and GPU offloading

Jan.ai

Open-source privacy-focused desktop assistant

Prompting & Sampling Tip: Runs optimally with standard chat templates (ChatML or Llama 3 format). A context window of 4,096 to 8,192 tokens balances memory usage and conversational memory.

GGUF file size and memory by quantization

Compare real GGUF weight sizes, estimated KV cache and total memory for Q4, Q5, Q8 and every quantization published in this repository.

Quant.Bits QualityWeights KVTotal Speed~Verdict
TQ1_0 1.89 Very low 78.69 GB 1.44 GB 80.93 GB Insufficient
IQ1_S 2.17 Very low 90.5 GB 1.44 GB 92.74 GB Insufficient
IQ1_M 2.4 Very low 100.27 GB 1.44 GB 102.5 GB Insufficient
IQ2_XXS 2.59 Very low 107.95 GB 1.44 GB 110.19 GB Insufficient
IQ2_M 2.73 Low 114.03 GB 1.44 GB 116.27 GB Insufficient
Q2_K 2.93 Low 122.15 GB 1.44 GB 124.39 GB Insufficient
Q2_K_L 2.93 Low 122.32 GB 1.44 GB 124.56 GB Insufficient
Q2_K_XL 3.02 Low 125.91 GB 1.44 GB 128.15 GB Insufficient
IQ3_XXS 3.24 Low 135.15 GB 1.44 GB 137.39 GB Insufficient
Q3_K_S 3.46 Fair 144.38 GB 1.44 GB 146.61 GB Insufficient
Q3_K_XL 3.54 Fair 147.84 GB 1.44 GB 150.08 GB Insufficient
Q3_K_M 3.82 Fair 159.51 GB 1.44 GB 161.74 GB Insufficient
IQ4_XS 4.28 Good 178.41 GB 1.44 GB 180.64 GB Insufficient
IQ4_NL 4.52 Good 188.55 GB 1.44 GB 190.79 GB Insufficient
Q4_0 4.53 Good 189.09 GB 1.44 GB 191.33 GB Insufficient
Q4_K_S 4.55 Good 189.69 GB 1.44 GB 191.93 GB Insufficient
Q4_K_XL 4.57 Good 190.52 GB 1.44 GB 192.75 GB Insufficient
Q4_K_M 4.83 Good 201.59 GB 1.44 GB 203.83 GB Insufficient
Q4_1 5.01 Very good 209.18 GB 1.44 GB 211.42 GB Insufficient
Q5_K_S 5.51 Very good 230.04 GB 1.44 GB 232.28 GB Insufficient
Q5_K_XL 5.66 Very good 236.19 GB 1.44 GB 238.43 GB Insufficient
Q5_K_M 5.68 Very good 236.79 GB 1.44 GB 239.02 GB Insufficient
Q6_K 6.57 Excellent 274.16 GB 1.44 GB 276.39 GB Insufficient
Q6_K_XL 6.72 Excellent 280.42 GB 1.44 GB 282.66 GB Insufficient
Q8_0 8.51 Excellent 354.8 GB 1.44 GB 357.04 GB Insufficient
Q8_K_XL 8.81 Excellent 367.72 GB 1.44 GB 369.95 GB Insufficient
BF16 16.0 Excellent 667.61 GB 1.44 GB 669.84 GB Insufficient

KV cache computed from the model's exact architecture. Speed is a rough estimate bounded by memory bandwidth.

Frequently asked questions

What kind of model is unsloth/GLM-4.7-GGUF?

unsloth/GLM-4.7-GGUF is a language model with 358.34 billion parameters, based on the glm4moe architecture. It is released under the mit license and distributed as GGUF files for local inference.

Is unsloth/GLM-4.7-GGUF a Mixture-of-Experts (MoE) model?

Yes. unsloth/GLM-4.7-GGUF is a Mixture-of-Experts model with 160 experts, of which 8 are activated per token. That makes it faster than a dense model of the same size, but all 160 experts must be loaded into memory, so the VRAM/RAM it needs is driven by the total parameter count, not the active one.

Can I run unsloth/GLM-4.7-GGUF on an 8 GB GPU?

No. unsloth/GLM-4.7-GGUF does not fit on an 8 GB GPU, even with the smallest quantization and system RAM offloading.

Can I run unsloth/GLM-4.7-GGUF on a 16 GB GPU?

No. unsloth/GLM-4.7-GGUF does not fit on a 16 GB GPU, even with the smallest quantization and system RAM offloading.

Can I run unsloth/GLM-4.7-GGUF on a 24 GB GPU?

No. unsloth/GLM-4.7-GGUF does not fit on a 24 GB GPU, even with the smallest quantization and system RAM offloading.

What context length does unsloth/GLM-4.7-GGUF support?

unsloth/GLM-4.7-GGUF supports a native context length of up to 202,752 tokens. A longer context grows the KV cache, so it increases the memory needed to run the model.

What is the best quantization for unsloth/GLM-4.7-GGUF?

For unsloth/GLM-4.7-GGUF, a strong default is Q4_K_M, which needs about 203.83 GB and keeps most of the quality while roughly halving the memory versus 8-bit. With VRAM to spare, Q5_K_M or Q6_K add a little more quality; if you are tight on memory, a smaller quantization still runs. Pick the highest quantization that fits your VRAM.