---
base_model: 0xSero/qwen3-coder-next-56b-REAP
base_model_relation: quantized
language:
- en
license: apache-2.0
tags:
- quantized
- gguf
- qwen3
- moe
---

# Qwen3-Coder-Next 56B REAP - GGUF

Quantized GGUF versions of [0xSero/qwen3-coder-next-56b-REAP](https://huggingface.co/0xSero/qwen3-coder-next-56b-REAP).
These were generated using the default settings with `llama-quantize` (b8740).

## Quantizations provided

| File | Quantization | Size |
|------|-------------|------|
| `qwen3-coder-next-56b-REAP-Q4_K_M.gguf` | Q4_K_M | 34.4 GB |
| `qwen3-coder-next-56b-REAP-Q5_K_M.gguf` | Q5_K_M | 40.3 GB |
| `qwen3-coder-next-56b-REAP-Q6_K.gguf` | Q6_K | 46.5 GB |
| `qwen3-coder-next-56b-REAP-Q8_0.gguf` | Q8_0 | 60.2 GB |

## Perplexity test

I tested perplexity using `llama-perplexity` and Salesforce's [wikitext-2-raw-v1](https://huggingface.co/datasets/Salesforce/wikitext/tree/main/wikitext-2-raw-v1).

| File | Quantization | Ctx | PPL |
|------|--------------|-----|-----|
| `qwen3-coder-next-56b-REAP-Q4_K_M.gguf` | Q4_K_M | 512 | 15.3702 +/- 0.13301 |
| `qwen3-coder-next-56b-REAP-Q5_K_M.gguf` | Q5_K_M | 512 | 15.2810 +/- 0.13196 |
| `qwen3-coder-next-56b-REAP-Q6_K.gguf` | Q6_K | 512 | 15.1305 +/- 0.13011 |
| `qwen3-coder-next-56b-REAP-Q8_0.gguf` | Q8_0 | 512 | 15.1198 +/- 0.13009 |
| `qwen3-coder-next-56b-REAP-BF16.gguf` | BF16 | 512 | 15.1274 +/- 0.13022 |