Total runs:
291
Run Growth:
149
Growth Rate:
51.20%
Sponsored by
Agentkit
Thireus / GLM-4.6-THIREUS-IQ1_M-SPECIAL_SPLIT
Model's Last Updated: March 30 2026
Introduction of GLM-4.6-THIREUS-IQ1_M-SPECIAL_SPLIT
Model Details of GLM-4.6-THIREUS-IQ1_M-SPECIAL_SPLIT
⚠️ Compatibility Notice
Support for the GLM-4.6 family was recently merged into both llama.cpp and ik_llama.cpp , so you must update to their latest versions before using any GGUF files from this repo.
- llama.cpp — see merged PR ggml-org/llama.cpp#16359
- ik_llama.cpp — see merged PR ikawrakow/ik_llama.cpp#814
GLM-4.6
🤔 What is this HuggingFace repository about?
This repository provides GGUF-quantized tensors for the GLM-4.6 model (official repo: https://huggingface.co/zai-org/GLM-4.6 ). These GGUF shards are designed to be used with Thireus’ GGUF Tool Suite ( https://gguf.thireus.com ), a collection of tools that automatically finds the perplexity-optimal mix of quantizations for any given VRAM and RAM target. With the Tool Suite, you can generate and download custom quantization “recipes” effortlessly.
- 📖 Read more: https://github.com/Thireus/GGUF-Tool-Suite
- 🔍 Example quant mixes: https://github.com/Thireus/GGUF-Tool-Suite/tree/main/recipe_examples
- 🛠️ Create your own recipe: https://colab.research.google.com/github/Thireus/GGUF-Tool-Suite/blob/main/quant_recipe_pipeline.ipynb
- 📂 Browse available quant shards: https://huggingface.co/Thireus/collections
tl;dr: Expand the details section below
cd ~
# Make sure to install all ik_llama.cpp compilation dependencies...
apt install python3-dev python3-pip python3-venv python3-wheel python3-setuptools git acl netcat-openbsd cmake # pipx
# Obtain ik_llama's Thireus version - Windows builds available at https://github.com/Thireus/ik_llama.cpp/releases
git clone https://github.com/Thireus/ik_llama.cpp
cd ik_llama.cpp
git pull
# Build ik_llama.cpp
cmake -B build -DGGML_AVX=ON -DGGML_AVX2=ON -DLLAMA_CURL=OFF -DGGML_MAX_CONTEXTS=2048
cmake --build build --config Release -j16
cd ..
# Obtain Thireus' GGUF-Tool-Suite
git clone https://github.com/Thireus/GGUF-Tool-Suite
# Download model quant mix from recipe file:
cd GGUF-Tool-Suite
rm -f download.conf # Make sure to copy the relevant download.conf for the model before running quant_assign.py
cp -f models/GLM-4.6/download.conf . # Use the download.conf of the chosen model
mkdir -p kitchen && cd kitchen
../quant_downloader.sh ../recipe_examples/ik_harmonized_recipes/GLM-4.6.ROOT-4.1636bpw-3.2647ppl.173GB-GGUF_12GB-GPU_160GB-CPU.90e3c2f_1ac651c.recipe
# Other recipe examples can be found at https://github.com/Thireus/GGUF-Tool-Suite/tree/main/recipe_examples
# Launch ik_llama's llama-server:
ulimit -n 9999 # Lifts "too many open files" limitation on Linux
~/ik_llama.cpp/build/bin/llama-server \
-m GLM-4.6-THIREUS-BF16-SPECIAL_TENSOR-00001-of-01762.gguf \
-fa -fmoe -ctk f16 -c 4096 -ngl 99 \
-ot "blk\.([0-9]|[1-2][0-9]|3[0-6])\.ffn_.*=CUDA0" \
-ot "blk\.(37|38|39|[4-6][0-9]|7[0-2])\.ffn_.*=CUDA1" \
-ot "blk\.(7[3-9])\.ffn_.*=CUDA2" \
-ot "blk\.(8[0-9]|90|91|92)\.ffn_.*=CPU" \
-ot exps=CPU -b 2048 -ub 1024 --warmup-batch --no-mmap --threads 36 \
--main-gpu 0
❓ Why does this Tool Suite exist?
-
Compatibility & Speed
–
unsloth
’s dynamic quants may not always work optimally with
ik_llama.cpp. - Custom Rig Fit – No off-the-shelf GGUF model perfectly matched my VRAM/RAM setup, so I built a way to tailor models and leverage extra VRAM/RAM to reduce perplexity.
- Automated PPL-Optimal Quantization – To my knowledge, there was no open source flexible, automated method to minimize perplexity for any bits-per-weight (bpw) target—so I created one with excellent results!
📊 How does it compare to other GGUFs?
Here’s how GLM-4.6 quantized with Thireus’ GGUF Tool Suite stacks up against other quantizers (lower perplexity = better at equal or lower bpw):
Note: The
recipe_examplesfiles illustrate good recipes. The Tool Suite computes the optimal ppl/bpw curve for you — just specify your target RAM, VRAM, and quant types, andquant_assign.pyfinds the best mix.
More perplexity/bpw graphs for other supported models: https://github.com/Thireus/GGUF-Tool-Suite/tree/main/ppl_graphs
All PPL benchmarks are computed with the parameters
-ctk f16 -c 512 -b 4096 -ub 4096
. Changing any of these parameters will alter the PPL. In particular, reducing
-b 4096 -ub 4096
increases the PPL, while increasing them decreases the PPL.
🚀 How do I get started?
Check out the GGUF Tool Suite README — focus on these sections:
-
⚠️
Requirements
– Which
ik_llama.cpp(orllama.cpp) version to use and how to compile.- Windows binaries (no patching needed) at: https://github.com/Thireus/ik_llama.cpp/releases
-
📥
Download Model Shards
– Use
quant_downloader.shto fetch GGUF shards from any recipe.- Recipe examples: https://github.com/Thireus/GGUF-Tool-Suite/tree/main/recipe_examples
-
🧠
Run a Downloaded Model
– Sample usage with
llama-cli. - 🛠️ Generate a Custom Recipe – Produce recipes tailored to your VRAM/RAM target usage for optimum perplexity.
✅ Supported Models
Supported models are listed under
models/
in the
Tool Suite Github repo
. Presence of
ppl_results.csv
indicates official support and compatibility with
quant_assign.py
.
🤷♂️ Will I release baked dynamic quant GGUFs?
No, because I believe in
tailored quantization
for each user’s hardware. If you prefer ready-made shards, you are welcome to merge them via
llama-gguf-split --merge
, or request someone to publish them, or rely on generic GGUF dynamic quants such as
unsloth
's.
Instead, I prefer to share examples of recipes so users can see exactly how they were produced (command included inside these recipe files) and tweak them for their own rigs. The
quant_downloader.sh
script handles automatic fetching and verification of each shard. Note that recipes provided by
Ubergarm
on his model cards are also compatible with
quant_downloader.sh
.
Users who don’t trust the GGUF shards on HuggingFace can also quantize their own by passing recipe lines to
llama-quantize --custom-q
(
see example
). Run
llama-quantize --help
to list compatible quants for
quant_assign.py
. This approach is especially useful if you prefer
llama.cpp
over
ik_llama.cpp
.
📦 What’s in this repository?
- 00001 GGUF header shard – Contains metadata (tokens, chat template, tensor count, etc.). This metadata can be explored directly from the HuggingFace web interface after clicking on that shard.
-
Tensor shards
– Each shard holds one tensor; see
tensors.mapfor names, quant types, sizes, SHA-256 hash, shard IDs, etc. -
GPG-signed files
–
tensors.mapand header shard are signed with the key in trusted-keys.asc for tamper detection. - Security note – Some papers about various ways to attack GGUFs and LLMs are available online, such as https://arxiv.org/abs/2505.23786 , and there are also more classic security exploits like CVE-2024-23496 and CVE-2024-25664 through CVE-2024-25668. Only use GGUFs from reputable, trusted authors—or alternatively self-quantize—to avoid potential exploits.
💡 Pro Tips
You can easily download the BF16 model version to quantize your own shards:
mkdir kitchen
echo '.*=bf16' > kitchen/bf16.recipe
cd kitchen
../quant_downloader.sh bf16.recipe
Enjoy optimized quantization! 🎉
Runs of Thireus GLM-4.6-THIREUS-IQ1_M-SPECIAL_SPLIT on huggingface.co
277
Total runs
0
24-hour runs
0
3-day runs
0
7-day runs
143
30-day runs
More Information About GLM-4.6-THIREUS-IQ1_M-SPECIAL_SPLIT huggingface.co Model
Url of GLM-4.6-THIREUS-IQ1_M-SPECIAL_SPLIT
GLM-4.6-THIREUS-IQ1_M-SPECIAL_SPLIT huggingface.co Url
Provider of GLM-4.6-THIREUS-IQ1_M-SPECIAL_SPLIT huggingface.co
Other API from Thireus
Updated:March 30 2026
Total runs:
286
Run Growth:
142
Growth Rate:
49.65%
Total runs:
283
Run Growth:
146
Growth Rate:
51.59%
Total runs:
281
Run Growth:
140
Growth Rate:
49.82%
Total runs:
281
Run Growth:
143
Growth Rate:
50.89%
Total runs:
279
Run Growth:
137
Growth Rate:
49.10%
Total runs:
277
Run Growth:
145
Growth Rate:
52.35%
Total runs:
275
Run Growth:
138
Growth Rate:
50.18%
Total runs:
274
Run Growth:
0
Growth Rate:
0.00%
Total runs:
272
Run Growth:
143
Growth Rate:
52.57%
Total runs:
271
Run Growth:
127
Growth Rate:
46.86%
Total runs:
270
Run Growth:
134
Growth Rate:
49.63%
Total runs:
270
Run Growth:
131
Growth Rate:
48.52%
Total runs:
269
Run Growth:
135
Growth Rate:
50.19%
Total runs:
266
Run Growth:
0
Growth Rate:
0.00%
Total runs:
265
Run Growth:
0
Growth Rate:
0.00%
Total runs:
263
Run Growth:
131
Growth Rate:
49.81%
Total runs:
199
Run Growth:
199
Growth Rate:
100.00%
Total runs:
182
Run Growth:
-2
Growth Rate:
-1.10%
Total runs:
179
Run Growth:
159
Growth Rate:
88.83%
Total runs:
173
Run Growth:
0
Growth Rate:
0.00%
Total runs:
170
Run Growth:
5
Growth Rate:
2.94%
Total runs:
170
Run Growth:
170
Growth Rate:
100.00%
Total runs:
159
Run Growth:
24
Growth Rate:
15.09%
Total runs:
156
Run Growth:
78
Growth Rate:
50.00%
Total runs:
155
Run Growth:
-165
Growth Rate:
-106.45%
Total runs:
149
Run Growth:
19
Growth Rate:
12.75%
Total runs:
144
Run Growth:
5
Growth Rate:
3.47%
Total runs:
142
Run Growth:
-2
Growth Rate:
-1.41%
Total runs:
140
Run Growth:
7
Growth Rate:
5.00%
Total runs:
140
Run Growth:
0
Growth Rate:
0.00%
Total runs:
140
Run Growth:
3
Growth Rate:
2.14%
Total runs:
139
Run Growth:
0
Growth Rate:
0.00%
Total runs:
137
Run Growth:
-7
Growth Rate:
-5.11%
Total runs:
137
Run Growth:
-11
Growth Rate:
-8.03%
Total runs:
137
Run Growth:
-6
Growth Rate:
-4.38%
Total runs:
136
Run Growth:
-2
Growth Rate:
-1.47%
Total runs:
136
Run Growth:
2
Growth Rate:
1.47%
Total runs:
136
Run Growth:
-4
Growth Rate:
-2.94%
Total runs:
135
Run Growth:
-12
Growth Rate:
-8.89%
Total runs:
135
Run Growth:
-5
Growth Rate:
-3.70%
Total runs:
135
Run Growth:
-4
Growth Rate:
-2.96%
Total runs:
135
Run Growth:
-4
Growth Rate:
-2.96%
Total runs:
135
Run Growth:
-5
Growth Rate:
-3.70%
Total runs:
132
Run Growth:
-7
Growth Rate:
-5.30%
Total runs:
132
Run Growth:
-3
Growth Rate:
-2.27%
Total runs:
131
Run Growth:
-5
Growth Rate:
-3.82%
Total runs:
84
Run Growth:
-84
Growth Rate:
-100.00%
Total runs:
78
Run Growth:
35
Growth Rate:
44.87%
Total runs:
76
Run Growth:
9
Growth Rate:
11.84%
Total runs:
73
Run Growth:
-201
Growth Rate:
-275.34%
Total runs:
73
Run Growth:
68
Growth Rate:
93.15%
Total runs:
72
Run Growth:
71
Growth Rate:
98.61%
Total runs:
62
Run Growth:
-9
Growth Rate:
-14.52%
Total runs:
60
Run Growth:
60
Growth Rate:
100.00%
Total runs:
51
Run Growth:
41
Growth Rate:
80.39%
Total runs:
48
Run Growth:
-3
Growth Rate:
-6.25%
Total runs:
47
Run Growth:
0
Growth Rate:
0.00%
Total runs:
47
Run Growth:
5
Growth Rate:
10.64%
Total runs:
44
Run Growth:
10
Growth Rate:
22.73%
Total runs:
42
Run Growth:
34
Growth Rate:
85.00%
Total runs:
41
Run Growth:
9
Growth Rate:
21.95%
Total runs:
41
Run Growth:
41
Growth Rate:
100.00%
Total runs:
40
Run Growth:
40
Growth Rate:
100.00%
Total runs:
39
Run Growth:
0
Growth Rate:
0.00%
Total runs:
38
Run Growth:
-18
Growth Rate:
-47.37%
Total runs:
38
Run Growth:
38
Growth Rate:
100.00%
Total runs:
37
Run Growth:
37
Growth Rate:
100.00%
Total runs:
37
Run Growth:
15
Growth Rate:
42.86%
Total runs:
36
Run Growth:
0
Growth Rate:
0.00%
Total runs:
36
Run Growth:
36
Growth Rate:
100.00%
Total runs:
35
Run Growth:
35
Growth Rate:
100.00%
Total runs:
35
Run Growth:
35
Growth Rate:
100.00%
Total runs:
34
Run Growth:
24
Growth Rate:
70.59%
Total runs:
32
Run Growth:
-18
Growth Rate:
-56.25%
Total runs:
32
Run Growth:
32
Growth Rate:
100.00%
Total runs:
31
Run Growth:
31
Growth Rate:
100.00%
Total runs:
30
Run Growth:
11
Growth Rate:
36.67%
Total runs:
30
Run Growth:
3
Growth Rate:
10.00%
Total runs:
29
Run Growth:
29
Growth Rate:
100.00%
Total runs:
28
Run Growth:
3
Growth Rate:
10.71%
Total runs:
28
Run Growth:
3
Growth Rate:
10.71%
Total runs:
28
Run Growth:
28
Growth Rate:
100.00%
Total runs:
28
Run Growth:
24
Growth Rate:
85.71%
Total runs:
28
Run Growth:
28
Growth Rate:
100.00%
Total runs:
28
Run Growth:
24
Growth Rate:
85.71%
Total runs:
28
Run Growth:
28
Growth Rate:
100.00%
Total runs:
28
Run Growth:
14
Growth Rate:
50.00%
Total runs:
27
Run Growth:
12
Growth Rate:
44.44%
Total runs:
27
Run Growth:
0
Growth Rate:
0.00%
Total runs:
27
Run Growth:
27
Growth Rate:
100.00%
Total runs:
27
Run Growth:
27
Growth Rate:
100.00%
Total runs:
26
Run Growth:
15
Growth Rate:
88.24%
Total runs:
25
Run Growth:
25
Growth Rate:
100.00%
Total runs:
25
Run Growth:
25
Growth Rate:
100.00%
Total runs:
25
Run Growth:
25
Growth Rate:
100.00%
Total runs:
24
Run Growth:
0
Growth Rate:
0.00%
Total runs:
23
Run Growth:
14
Growth Rate:
60.87%
Total runs:
23
Run Growth:
-52
Growth Rate:
-226.09%
Total runs:
23
Run Growth:
23
Growth Rate:
100.00%