Instructions to use bartowski/nex-agi_Nex-N2.5-mini-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use bartowski/nex-agi_Nex-N2.5-mini-GGUF with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf bartowski/nex-agi_Nex-N2.5-mini-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf bartowski/nex-agi_Nex-N2.5-mini-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf bartowski/nex-agi_Nex-N2.5-mini-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf bartowski/nex-agi_Nex-N2.5-mini-GGUF:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf bartowski/nex-agi_Nex-N2.5-mini-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf bartowski/nex-agi_Nex-N2.5-mini-GGUF:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf bartowski/nex-agi_Nex-N2.5-mini-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf bartowski/nex-agi_Nex-N2.5-mini-GGUF:Q4_K_M

Use Docker

docker model run hf.co/bartowski/nex-agi_Nex-N2.5-mini-GGUF:Q4_K_M

LM Studio
Jan

vLLM

How to use bartowski/nex-agi_Nex-N2.5-mini-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "bartowski/nex-agi_Nex-N2.5-mini-GGUF"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "bartowski/nex-agi_Nex-N2.5-mini-GGUF",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/bartowski/nex-agi_Nex-N2.5-mini-GGUF:Q4_K_M

Ollama
How to use bartowski/nex-agi_Nex-N2.5-mini-GGUF with Ollama:
```
ollama run hf.co/bartowski/nex-agi_Nex-N2.5-mini-GGUF:Q4_K_M
```
Unsloth Desktop

How to use bartowski/nex-agi_Nex-N2.5-mini-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf bartowski/nex-agi_Nex-N2.5-mini-GGUF:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "bartowski/nex-agi_Nex-N2.5-mini-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use bartowski/nex-agi_Nex-N2.5-mini-GGUF with Docker Model Runner:
```
docker model run hf.co/bartowski/nex-agi_Nex-N2.5-mini-GGUF:Q4_K_M
```

Lemonade

How to use bartowski/nex-agi_Nex-N2.5-mini-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull bartowski/nex-agi_Nex-N2.5-mini-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.nex-agi_Nex-N2.5-mini-GGUF-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use bartowski/nex-agi_Nex-N2.5-mini-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf bartowski/nex-agi_Nex-N2.5-mini-GGUF:Q4_K_M

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default bartowski/nex-agi_Nex-N2.5-mini-GGUF:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use bartowski/nex-agi_Nex-N2.5-mini-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf bartowski/nex-agi_Nex-N2.5-mini-GGUF:Q4_K_M

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "bartowski/nex-agi_Nex-N2.5-mini-GGUF:Q4_K_M" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

Llamacpp imatrix Quantizations of Nex-N2.5-mini by nex-agi

Using llama.cpp release b10896 for quantization.

Original model: https://huggingface.co/nex-agi/Nex-N2.5-mini

Model details:

Parameter count: 35B
Input support: text, image (with mmproj file) - details
Speculative decoding: no
imatrix: yes - details

How to run

Prompt format

<|im_start|>system
{system_prompt}<|im_end|>
<|im_start|>user
{prompt}<|im_end|>
<|im_start|>assistant
<think>

Don't know which to choose? Grab Q4_K_M (22.32GB) - usually a good mix of size and performance. Download instructions available here

Available files:

Filename	Quant type	File Size	Split	Description
nex-agi_Nex-N2.5-mini-bf16.gguf	bf16	69.38GB	true	Full BF16 weights.
nex-agi_Nex-N2.5-mini-Q8_0.gguf	Q8_0	36.91GB	false	Extremely high quality, generally unneeded but max available quant.
nex-agi_Nex-N2.5-mini-Q6_K_L.gguf	Q6_K_L	32.30GB	false	The large size of Q6_K, about halfway to Q8_0 in size. Very high quality, near perfect, recommended.
nex-agi_Nex-N2.5-mini-Q6_K.gguf	Q6_K	29.37GB	false	Very high quality, near perfect, recommended.
nex-agi_Nex-N2.5-mini-Q5_K_M.gguf	Q5_K_M	26.98GB	false	High quality, recommended.
nex-agi_Nex-N2.5-mini-Q5_K_S.gguf	Q5_K_S	25.14GB	false	High quality, recommended.
nex-agi_Nex-N2.5-mini-Q4_K_L.gguf	Q4_K_L	24.11GB	false	The large size of Q4_K, between Q4_K_M and Q5_K_S: more of the most sensitive weights kept at higher precision, recommended.
nex-agi_Nex-N2.5-mini-Q4_K_M.gguf	Q4_K_M	22.32GB	false	Good quality, default size for most use cases, recommended.
nex-agi_Nex-N2.5-mini-IQ4_NL.gguf	IQ4_NL	22.32GB	false	Similar to IQ4_XS, but slightly larger.
nex-agi_Nex-N2.5-mini-Q4_1.gguf	Q4_1	21.97GB	false	Legacy format, similar performance to Q4_K_S but with improved tokens/watt on Apple silicon.
nex-agi_Nex-N2.5-mini-Q4_K_S.gguf	Q4_K_S	20.95GB	false	Slightly lower quality with more space savings, recommended.
nex-agi_Nex-N2.5-mini-Q4_0.gguf	Q4_0	19.94GB	false	Legacy format, kept for compatibility with older tools.
nex-agi_Nex-N2.5-mini-IQ4_XS.gguf	IQ4_XS	19.89GB	false	Decent quality, smaller than Q4_K_S with similar performance, recommended.
nex-agi_Nex-N2.5-mini-IQ3_M.gguf	IQ3_M	18.96GB	false	Medium-low quality, new method with decent performance comparable to Q3_K_M.
nex-agi_Nex-N2.5-mini-Q3_K_L.gguf	Q3_K_L	18.27GB	false	Lower quality but usable, good for low RAM availability.
nex-agi_Nex-N2.5-mini-Q3_K_M.gguf	Q3_K_M	17.34GB	false	Low quality.
nex-agi_Nex-N2.5-mini-IQ3_XS.gguf	IQ3_XS	16.32GB	false	Lower quality, new method with decent performance, slightly better than Q3_K_S.
nex-agi_Nex-N2.5-mini-Q3_K_S.gguf	Q3_K_S	16.32GB	false	Low quality, not recommended.
nex-agi_Nex-N2.5-mini-IQ3_XXS.gguf	IQ3_XXS	15.58GB	false	Lower quality, new method with decent performance, comparable to Q3 quants.
nex-agi_Nex-N2.5-mini-Q2_K.gguf	Q2_K	13.84GB	false	Very low quality but surprisingly usable.
nex-agi_Nex-N2.5-mini-IQ2_M.gguf	IQ2_M	13.23GB	false	Relatively low quality, uses SOTA techniques to be surprisingly usable.
nex-agi_Nex-N2.5-mini-IQ2_S.gguf	IQ2_S	11.98GB	false	Low quality, uses SOTA techniques to be usable.
nex-agi_Nex-N2.5-mini-IQ2_XS.gguf	IQ2_XS	11.27GB	false	Low quality, uses SOTA techniques to be usable.
nex-agi_Nex-N2.5-mini-IQ2_XXS.gguf	IQ2_XXS	10.79GB	false	Very low quality, uses SOTA techniques to be usable.

Download a specific file:

hf download bartowski/nex-agi_Nex-N2.5-mini-GGUF --include "nex-agi_Nex-N2.5-mini-Q4_K_M.gguf" --local-dir ./

Downloading using the Hugging Face CLI

Click to view download instructions

First, make sure you have the Hugging Face CLI installed:

pip install -U "huggingface_hub[cli]"

Download a specific file:

hf download bartowski/nex-agi_Nex-N2.5-mini-GGUF --include "nex-agi_Nex-N2.5-mini-Q4_K_M.gguf" --local-dir ./

The files marked true in the Split column above are stored as multiple parts in a folder. To download all the parts to a local folder, run:

hf download bartowski/nex-agi_Nex-N2.5-mini-GGUF --include "nex-agi_Nex-N2.5-mini-bf16/*" --local-dir ./

You can either specify a new local-dir (nex-agi_Nex-N2.5-mini-bf16) or download them all in place (./)

How to run

These quants run with llama.cpp - installable in one line via llama.app:

curl -LsSf https://llama.app/install.sh | sh
llama-server -hf bartowski/nex-agi_Nex-N2.5-mini-GGUF:Q4_K_M

llama-server includes a built-in chat web UI, served at http://localhost:8080 by default.

These quants were made with llama.cpp release b10896 - if this model's architecture is newly supported, you'll need that release or newer to run them.

They also work in: LM Studio · koboldcpp · ramalama · Jan AI · Text Generation Web UI · LoLLMs · Atomic Chat

Multimodal

This model supports image input. Alongside the quants, this repo includes the multimodal projector files mmproj-nex-agi_Nex-N2.5-mini-f16.gguf and mmproj-nex-agi_Nex-N2.5-mini-bf16.gguf, which pair with any quant above.

llama.cpp downloads the mmproj automatically when using -hf as shown above; if you're loading files manually, pass it with --mmproj.

Per-tensor layouts

Some of these files were built with a layout computed for this model instead of llama.cpp's standard one-size-fits-all rules. A Q4_K_M is still mostly Q4_K; the extra precision goes to the weights this particular model is most sensitive to. The S, M or L in a name says how much of the model stays at the base precision: about 90 % for S, 70 % for M and 50 % for L. An _L name is simply the large size of its family. Q4_K_L is to Q4_K_M what Q4_K_M is to Q4_K_S, and Q6_K_L sits about halfway between Q6_K and Q8_0. In earlier releases an _L name meant the embedding and output weights were kept at Q8_0; in these files it means the larger size of the base type. There is no size target, so each file's bits per weight is reported rather than promised.

The layout each of these files was built with is published in the layouts/ folder: <file>.tensor-types.txt is the exact --tensor-type-file given to llama-quantize, and <file>.layout.json records how it was computed, including the generator version, the llama.cpp release and the commit, so any of them can be rebuilt.

Checked on this model before any of these files were released: Q4_K_M reached 0.85×, Q3_K_M 0.80× and IQ2_XXS 0.87× the KL divergence of the standard layout at the same file size.

Layout details

Files built from a computed layout:

Quant	Size	Body bits/weight	File bits/weight	Body kept at base type
Q6_K_L	32.30GB	7.40	7.36	50 %
Q6_K	29.37GB	6.71	6.69	91 %
Q5_K_M	26.98GB	6.14	6.15	70 %
Q5_K_S	25.14GB	5.70	5.73	90 %
Q4_K_L	24.11GB	5.45	5.49	50 %
Q4_K_M	22.32GB	5.03	5.09	70 %
IQ4_NL	22.32GB	5.03	5.09	70 %
Q4_K_S	20.95GB	4.70	4.77	90 %
IQ4_XS	19.89GB	4.45	4.53	90 %
IQ3_M	18.96GB	4.23	4.32	50 %
Q3_K_L	18.27GB	4.06	4.16	50 %
Q3_K_M	17.34GB	3.84	3.95	70 %
IQ3_XS	16.32GB	3.60	3.72	90 %
Q3_K_S	16.32GB	3.60	3.72	90 %
IQ3_XXS	15.58GB	3.42	3.55	70 %
Q2_K	13.84GB	3.05	3.15	70 %
IQ2_M	13.23GB	2.87	3.02	70 %
IQ2_S	11.98GB	2.61	2.73	70 %
IQ2_XS	11.27GB	2.44	2.57	90 %
IQ2_XXS	10.79GB	2.33	2.46	70 %

Checked on this model: the computed layout against the standard one, measured by KL divergence against the unquantized model. The ratio compares each computed file with the standard ladder read at that file's own size, so it can differ from the two KLD columns when the two files differ in size.

Quant	Computed layout KLD	Standard layout KLD	Ratio at equal size	Size vs standard file
Q4_K_M	0.0302 ± 0.0006	0.0415 ± 0.0006	0.85×	+4.3 %
Q3_K_M	0.0659 ± 0.0009	0.0991 ± 0.0014	0.80×	+6.9 %
IQ2_XXS	0.2401 ± 0.0026	0.3331 ± 0.0033	0.87×	+10.4 %

How it works: the base type is a floor for every body tensor and a fixed share of the body bytes stays at it (90 % for S, 70 % for M, 50 % for L); the remaining bytes go where a cross-model sensitivity prior, measured by KL divergence against the unquantized model, says they buy the most quality. The embedding and output tensors are sized by their share of the file: a small table is kept at Q8_0, a large one follows the file's bitrate. A K-quant and the IQ quant with the same base bitrate (Q3_K_S and IQ3_XS, Q3_K_M and IQ3_S, Q3_K_L and IQ3_M) come out at about the same size; the IQ file is the GPU-oriented twin.

imatrix

All quants made using imatrix option, with a calibration corpus rendered through this model's own chat template. The corpus pairs plain prose with tool-calling and reasoning conversations (corpus source data), encoded exactly as this model sees them at inference and processed with --parse-special, so chat-format special tokens contribute to the importance matrix. The corpus rendered for this model is included in this repo: nex-agi_Nex-N2.5-mini-calibration-v6.txt. The imatrix is available here: nex-agi_Nex-N2.5-mini-imatrix.gguf.

Calibration render details

{
  "generator": "auto_quant_v2 calibration renderer",
  "recipe": "calibration-v6",
  "model": "Nex-N2.5-mini",
  "encoder": "chat_template",
  "chunk_size": 512,
  "prose_chunks": 214,
  "tool_chunks": 359,
  "total_chunks": 573,
  "tool_chunk_fraction": 0.627,
  "n_conversations": 137,
  "extension_convs_used": 0,
  "conversation_token_lengths": [
    566,
    1629,
    1259,
    1521,
    1147,
    1366,
    3182,
    828,
    1229,
    1424,
    1069,
    2098,
    883,
    1255,
    2791,
    1275,
    1122,
    988,
    739,
    720,
    1364,
    1060,
    1409,
    1204,
    1874,
    1500,
    1656,
    910,
    1417,
    1648,
    1571,
    1222,
    1270,
    1039,
    1018,
    1714,
    1653,
    1177,
    479,
    1912,
    1413,
    1128,
    1400,
    2004,
    2092,
    1317,
    1611,
    909,
    2958,
    1104,
    2870,
    789,
    1026,
    972,
    905,
    700,
    2492,
    917,
    1150,
    1089,
    1233,
    1173,
    928,
    1221,
    1185,
    1590,
    915,
    1554,
    2128,
    843,
    323,
    1124,
    3337,
    2838,
    711,
    944,
    1034,
    1075,
    1300,
    1084,
    1153,
    797,
    1212,
    1073,
    1293,
    1548,
    1405,
    2068,
    875,
    660,
    2768,
    640,
    1405,
    1677,
    1949,
    1205,
    645,
    1358,
    1126,
    1731,
    1846,
    1717,
    819,
    1017,
    1039,
    2858,
    742,
    736,
    765,
    1414,
    1054,
    1579,
    770,
    350,
    319,
    2628,
    1020,
    1141,
    1832,
    2041,
    2697,
    2679,
    818,
    982,
    842,
    961,
    1241,
    1001,
    844,
    1359,
    832,
    712,
    1745,
    1029,
    919,
    1318,
    1481
  ],
  "warnings": []
}

ARM/AVX information

llama.cpp automatically "repacks" weights into an interleaved layout at load time for faster inference on ARM and AVX machines - details in this PR. This once required downloading special Q4_0_4_4/4_8/8_8 files; those are long gone. Online repacking now covers Q4_0, IQ4_NL, and most K-quants, so no special quant choice is needed for CPU inference.

Which file should I choose?

Click here for details

An older (early 2024) but still useful write-up with charts comparing quant performances is provided by Artefact2 here

The first thing to figure out is how big a model you can run. To do this, you'll need to figure out how much RAM and/or VRAM you have.

If you want your model running as FAST as possible, you'll want to fit the whole thing on your GPU's VRAM. Aim for a quant with a file size 1-2GB smaller than your GPU's total VRAM.

If you want the absolute maximum quality, add both your system RAM and your GPU's VRAM together, then similarly grab a quant with a file size 1-2GB Smaller than that total.

Hugging Face can also do this math for you: add your hardware in your Local Apps settings and the model page will show which files fit.

Next, you'll need to decide if you want to use an 'I-quant' or a 'K-quant'.

If you don't want to think too much, grab one of the K-quants. These are in format 'QX_K_X', like Q5_K_M.

If you want to get more into the weeds, you can check out this extremely useful feature chart:

llama.cpp feature matrix

But basically, if you're aiming for below Q4, and you're running cuBLAS (Nvidia) or rocBLAS (AMD), you should look towards the I-quants. These are in format IQX_X, like IQ3_M. These are newer and offer better performance for their size.

These I-quants can also be used on CPU, but will be slower than their K-quant equivalent, so speed vs performance is a tradeoff you'll have to decide.