---
license: cc-by-nc-4.0
tags:
- moe
- frankenmoe
- merge
- mergekit
- lazymergekit
base_model:
- mlabonne/AlphaMonarch-7B
- beowolx/CodeNinja-1.0-OpenChat-7B
- SanjiWatsuki/Kunoichi-DPO-v2-7B
- mlabonne/NeuralDaredevil-7B
quantized_by: bartowski
pipeline_tag: text-generation
---

## Llamacpp Quantizations of Beyonder-4x7B-v3

Using <a href="https://github.com/ggerganov/llama.cpp/">llama.cpp</a> release <a href="https://github.com/ggerganov/llama.cpp/releases/tag/b2440">b2440</a> for quantization.

Original model: https://huggingface.co/mlabonne/Beyonder-4x7B-v3

Download a file (not the whole branch) from below:

| Filename | Quant type | File Size | Description |
| -------- | ---------- | --------- | ----------- |
| [Beyonder-4x7B-v3-Q8_0.gguf](https://huggingface.co/bartowski/Beyonder-4x7B-v3-GGUF/blob/main/Beyonder-4x7B-v3-Q8_0.gguf) | Q8_0 | 25.66GB | Extremely high quality, generally unneeded but max available quant. |
| [Beyonder-4x7B-v3-Q6_K.gguf](https://huggingface.co/bartowski/Beyonder-4x7B-v3-GGUF/blob/main/Beyonder-4x7B-v3-Q6_K.gguf) | Q6_K | 19.81GB | Very high quality, near perfect, *recommended*. |
| [Beyonder-4x7B-v3-Q5_K_M.gguf](https://huggingface.co/bartowski/Beyonder-4x7B-v3-GGUF/blob/main/Beyonder-4x7B-v3-Q5_K_M.gguf) | Q5_K_M | 17.13GB | High quality, very usable. |
| [Beyonder-4x7B-v3-Q5_K_S.gguf](https://huggingface.co/bartowski/Beyonder-4x7B-v3-GGUF/blob/main/Beyonder-4x7B-v3-Q5_K_S.gguf) | Q5_K_S | 16.62GB | High quality, very usable. |
| [Beyonder-4x7B-v3-Q5_0.gguf](https://huggingface.co/bartowski/Beyonder-4x7B-v3-GGUF/blob/main/Beyonder-4x7B-v3-Q5_0.gguf) | Q5_0 | 16.62GB | High quality, older format, generally not recommended. |
| [Beyonder-4x7B-v3-Q4_K_M.gguf](https://huggingface.co/bartowski/Beyonder-4x7B-v3-GGUF/blob/main/Beyonder-4x7B-v3-Q4_K_M.gguf) | Q4_K_M | 14.60GB | Good quality, similar to 4.25 bpw. |
| [Beyonder-4x7B-v3-Q4_K_S.gguf](https://huggingface.co/bartowski/Beyonder-4x7B-v3-GGUF/blob/main/Beyonder-4x7B-v3-Q4_K_S.gguf) | Q4_K_S | 13.74GB | Slightly lower quality with small space savings. |
| [Beyonder-4x7B-v3-IQ4_NL.gguf](https://huggingface.co/bartowski/Beyonder-4x7B-v3-GGUF/blob/main/Beyonder-4x7B-v3-IQ4_NL.gguf) | IQ4_NL | 13.75GB | Good quality, similar to Q4_K_S, new method of quanting, |
| [Beyonder-4x7B-v3-IQ4_XS.gguf](https://huggingface.co/bartowski/Beyonder-4x7B-v3-GGUF/blob/main/Beyonder-4x7B-v3-IQ4_XS.gguf) | IQ4_XS | 13.03GB | Decent quality, new method with similar performance to Q4. |
| [Beyonder-4x7B-v3-Q4_0.gguf](https://huggingface.co/bartowski/Beyonder-4x7B-v3-GGUF/blob/main/Beyonder-4x7B-v3-Q4_0.gguf) | Q4_0 | 13.62GB | Decent quality, older format, generally not recommended. |
| [Beyonder-4x7B-v3-IQ3_M.gguf](https://huggingface.co/bartowski/Beyonder-4x7B-v3-GGUF/blob/main/Beyonder-4x7B-v3-IQ3_M.gguf) | IQ3_M | 10.64GB | Medium-low quality, new method with decent performance. |
| [Beyonder-4x7B-v3-IQ3_S.gguf](https://huggingface.co/bartowski/Beyonder-4x7B-v3-GGUF/blob/main/Beyonder-4x7B-v3-IQ3_S.gguf) | IQ3_S | 10.45GB | Lower quality, new method with decent performance, recommended over Q3 quants. |
| [Beyonder-4x7B-v3-Q3_K_L.gguf](https://huggingface.co/bartowski/Beyonder-4x7B-v3-GGUF/blob/main/Beyonder-4x7B-v3-Q3_K_L.gguf) | Q3_K_L | 12.54GB | Lower quality but usable, good for low RAM availability. |
| [Beyonder-4x7B-v3-Q3_K_M.gguf](https://huggingface.co/bartowski/Beyonder-4x7B-v3-GGUF/blob/main/Beyonder-4x7B-v3-Q3_K_M.gguf) | Q3_K_M | 11.57GB | Even lower quality. |
| [Beyonder-4x7B-v3-Q3_K_S.gguf](https://huggingface.co/bartowski/Beyonder-4x7B-v3-GGUF/blob/main/Beyonder-4x7B-v3-Q3_K_S.gguf) | Q3_K_S | 10.43GB | Low quality, not recommended. |
| [Beyonder-4x7B-v3-Q2_K.gguf](https://huggingface.co/bartowski/Beyonder-4x7B-v3-GGUF/blob/main/Beyonder-4x7B-v3-Q2_K.gguf) | Q2_K | 8.84GB | Extremely low quality, *not* recommended.

Want to support my work? Visit my ko-fi page here: https://ko-fi.com/bartowski