---
quantized_by: bartowski
pipeline_tag: image-text-to-text
license_link: https://huggingface.co/ornith-ai/Ornith-1.5-397B/blob/main/LICENSE
base_model_relation: quantized
base_model: ornith-ai/Ornith-1.5-397B
license: mit
---
## Llamacpp imatrix Quantizations of Ornith-1.5-397B by ornith-ai
Using llama.cpp release b10472 for quantization.
Original model: https://huggingface.co/ornith-ai/Ornith-1.5-397B
**Model details:**
- Parameter count: 397B
- Input support: text, image (with mmproj file) - [details](#multimodal)
- Speculative decoding: no
- imatrix: yes - [details](#imatrix)
[How to run](#how-to-run)
## Prompt format
```
<|im_start|>system
{system_prompt}<|im_end|>
<|im_start|>user
{prompt}<|im_end|>
<|im_start|>assistant
```
**Don't know which to choose?** Grab [Q4_K_M](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/tree/main/Ornith-1.5-397B-Q4_K_M) (241.81GB) - usually a good mix of size and performance. Download instructions available [here](#downloading-using-the-hugging-face-cli)
## Available files:
| Filename | Quant type | File Size | Split | Description |
| -------- | ---------- | --------- | ----- | ----------- |
| [Ornith-1.5-397B-Q8_0.gguf](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/tree/main/Ornith-1.5-397B-Q8_0) | Q8_0 | 421.58GB | true | Extremely high quality, generally unneeded but max available quant. |
| [Ornith-1.5-397B-Q6_K.gguf](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/tree/main/Ornith-1.5-397B-Q6_K) | Q6_K | 342.39GB | true | Very high quality, near perfect, *recommended*. |
| [Ornith-1.5-397B-Q5_K_M.gguf](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/tree/main/Ornith-1.5-397B-Q5_K_M) | Q5_K_M | 283.26GB | true | High quality, *recommended*. |
| [Ornith-1.5-397B-Q5_K_S.gguf](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/tree/main/Ornith-1.5-397B-Q5_K_S) | Q5_K_S | 273.99GB | true | High quality, *recommended*. |
| [Ornith-1.5-397B-Q4_1.gguf](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/tree/main/Ornith-1.5-397B-Q4_1) | Q4_1 | 249.10GB | true | Legacy format, similar performance to Q4_K_S but with improved tokens/watt on Apple silicon. |
| [Ornith-1.5-397B-Q4_K_M.gguf](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/tree/main/Ornith-1.5-397B-Q4_K_M) | Q4_K_M | 241.81GB | true | Good quality, default size for most use cases, *recommended*. |
| [Ornith-1.5-397B-Q4_K_S.gguf](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/tree/main/Ornith-1.5-397B-Q4_K_S) | Q4_K_S | 232.84GB | true | Slightly lower quality with more space savings, *recommended*. |
| [Ornith-1.5-397B-Q4_0.gguf](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/tree/main/Ornith-1.5-397B-Q4_0) | Q4_0 | 225.44GB | true | Legacy format, kept for compatibility with older tools. |
| [Ornith-1.5-397B-IQ4_NL.gguf](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/tree/main/Ornith-1.5-397B-IQ4_NL) | IQ4_NL | 224.52GB | true | Similar to IQ4_XS, but slightly larger. |
| [Ornith-1.5-397B-IQ4_XS.gguf](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/tree/main/Ornith-1.5-397B-IQ4_XS) | IQ4_XS | 212.23GB | true | Decent quality, smaller than Q4_K_S with similar performance, *recommended*. |
| [Ornith-1.5-397B-Q3_K_XL.gguf](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/tree/main/Ornith-1.5-397B-Q3_K_XL) | Q3_K_XL | 190.37GB | true | Uses Q8_0 for embed and output weights. Lower quality but usable, good for low RAM availability. |
| [Ornith-1.5-397B-IQ3_M.gguf](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/tree/main/Ornith-1.5-397B-IQ3_M) | IQ3_M | 189.56GB | true | Medium-low quality, new method with decent performance comparable to Q3_K_M. |
| [Ornith-1.5-397B-Q3_K_L.gguf](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/tree/main/Ornith-1.5-397B-Q3_K_L) | Q3_K_L | 189.48GB | true | Lower quality but usable, good for low RAM availability. |
| [Ornith-1.5-397B-Q3_K_M.gguf](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/tree/main/Ornith-1.5-397B-Q3_K_M) | Q3_K_M | 181.49GB | true | Low quality. |
| [Ornith-1.5-397B-IQ3_XS.gguf](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/tree/main/Ornith-1.5-397B-IQ3_XS) | IQ3_XS | 181.44GB | true | Lower quality, new method with decent performance, slightly better than Q3_K_S. |
| [Ornith-1.5-397B-Q3_K_S.gguf](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/tree/main/Ornith-1.5-397B-Q3_K_S) | Q3_K_S | 172.93GB | true | Low quality, not recommended. |
| [Ornith-1.5-397B-IQ3_XXS.gguf](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/tree/main/Ornith-1.5-397B-IQ3_XXS) | IQ3_XXS | 166.08GB | true | Lower quality, new method with decent performance, comparable to Q3 quants. |
| [Ornith-1.5-397B-Q2_K_L.gguf](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/tree/main/Ornith-1.5-397B-Q2_K_L) | Q2_K_L | 140.49GB | true | Uses Q8_0 for embed and output weights. Very low quality but surprisingly usable. |
| [Ornith-1.5-397B-Q2_K.gguf](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/tree/main/Ornith-1.5-397B-Q2_K) | Q2_K | 139.50GB | true | Very low quality but surprisingly usable. |
| [Ornith-1.5-397B-IQ2_M.gguf](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/tree/main/Ornith-1.5-397B-IQ2_M) | IQ2_M | 133.13GB | true | Relatively low quality, uses SOTA techniques to be surprisingly usable. |
| [Ornith-1.5-397B-IQ2_S.gguf](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/tree/main/Ornith-1.5-397B-IQ2_S) | IQ2_S | 120.51GB | true | Low quality, uses SOTA techniques to be usable. |
| [Ornith-1.5-397B-IQ2_XS.gguf](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/tree/main/Ornith-1.5-397B-IQ2_XS) | IQ2_XS | 118.47GB | true | Low quality, uses SOTA techniques to be usable. |
| [Ornith-1.5-397B-IQ2_XXS.gguf](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/tree/main/Ornith-1.5-397B-IQ2_XXS) | IQ2_XXS | 106.31GB | true | Very low quality, uses SOTA techniques to be usable. |
| [Ornith-1.5-397B-IQ1_M.gguf](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/tree/main/Ornith-1.5-397B-IQ1_M) | IQ1_M | 91.36GB | true | Extremely low quality, *not* recommended. |
| [Ornith-1.5-397B-IQ1_S.gguf](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/tree/main/Ornith-1.5-397B-IQ1_S) | IQ1_S | 81.78GB | true | Extremely low quality, *not* recommended. |
Download a specific file:
```
hf download bartowski/Ornith-1.5-397B-GGUF --include "Ornith-1.5-397B-Q4_K_M/*" --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/Ornith-1.5-397B-GGUF --include "Ornith-1.5-397B-Q4_K_M/*" --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/Ornith-1.5-397B-GGUF --include "Ornith-1.5-397B-Q8_0/*" --local-dir ./
```
You can either specify a new local-dir (Ornith-1.5-397B-Q8_0) or download them all in place (./)
## How to run
These quants run with [llama.cpp](https://github.com/ggml-org/llama.cpp) - installable in one line via [llama.app](https://llama.app/):
```
curl -LsSf https://llama.app/install.sh | sh
llama-server -hf bartowski/Ornith-1.5-397B-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 b10472 - if this model's architecture is newly supported, you'll need that release or newer to run them.
They also work in: [LM Studio](https://lmstudio.ai/) · [koboldcpp](https://github.com/LostRuins/koboldcpp) · [ramalama](https://github.com/containers/ramalama) · [Jan AI](https://www.jan.ai/) · [Text Generation Web UI](https://github.com/oobabooga/text-generation-webui) · [LoLLMs](https://github.com/ParisNeo/lollms) · [Atomic Chat](https://atomic.chat/)
## Multimodal
This model supports image input. Alongside the quants, this repo includes the multimodal projector files [mmproj-Ornith-1.5-397B-f16.gguf](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/blob/main/mmproj-Ornith-1.5-397B-f16.gguf) and [mmproj-Ornith-1.5-397B-bf16.gguf](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/blob/main/mmproj-Ornith-1.5-397B-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`.
## 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](https://gist.github.com/bartowski1182/e26453c0404e24eb317543ec5360f87a)), 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: [Ornith-1.5-397B-calibration-v6.txt](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/blob/main/Ornith-1.5-397B-calibration-v6.txt). The imatrix is available here: [Ornith-1.5-397B-imatrix.gguf](https://huggingface.co/bartowski/Ornith-1.5-397B-GGUF/blob/main/Ornith-1.5-397B-imatrix.gguf).
Calibration render details
```json
{
"generator": "auto_quant_v2 calibration renderer",
"recipe": "calibration-v6",
"model": "Ornith-1.5-397B",
"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": []
}
```
## Embed/output weights
Some of these quants (Q3_K_XL, Q4_K_L etc) are the standard quantization method with the embeddings and output weights quantized to Q8_0 instead of what they would normally default to.
## 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](https://github.com/ggml-org/llama.cpp/pull/9921). 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](https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9)
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](https://huggingface.co/settings/local-apps) 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](https://github.com/ggml-org/llama.cpp/wiki/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.
## Credits
Thank you kalomaze and Dampf for assistance in creating the imatrix calibration dataset.
Thank you ZeroWw for the inspiration to experiment with embed/output.
Want to support my work? Visit my ko-fi page here: https://ko-fi.com/bartowski