---

base_model: Qwen/Qwen3.8-Flash-Next

base_model_relation: quantized

license: gemma

library_name: llama.cpp

pipeline_tag: text-generation

quantized_by: spiritfather

tags:

• gguf
• creative-writing
• llama.cpp

---

cyberneurova-Qwen3.8-Flash-Next — GGUF (static quants)

<p align="center">

🧬 Source model: <a href="https://huggingface.co/cyberneurova/cyberneurova-Qwen3.8-Flash-Next" target="_blank">cyberneurova/cyberneurova-Qwen3.8-Flash-Next</a>

<br>

📦 imatrix (i1) quants: <a href="https://huggingface.co/spiritfather/cyberneurova-Qwen3.8-Flash-Next-i1-GGUF" target="_blank">spiritfather/cyberneurova-Qwen3.8-Flash-Next-i1-GGUF</a>

<br>

🛠️ Runs on <a href="https://github.com/ggml-org/llama.cpp" target="_blank">llama.cpp</a> (master)

<br>

📊 Scored on <a href="https://caliperbench.com" target="_blank">CaliperBench</a>

<br>

💬 <a href="https://discord.gg/rhE9zdKyE4" target="_blank">Discord</a>

</p>

> [!Note]

> All credit for the model goes to cyberneurova, who produced cyberneurova-Qwen3.8-Flash-Next. These are quantizations only.

> [!Tip]

> Benchmarked on CaliperBench — a creative-writing benchmark scoring prose craft, roleplay and willingness rather than general intelligence. See this model's scores: caliperbench.com

Provided quants

| Quant | Size | Notes |

|:--|--:|:--|

| Q2_K | 80.4 GB | |

| Q3_K_S | 88.8 GB | |

| Q3_K_M | 94.2 GB | lower quality |

| Q4_K_M | 119.2 GB | fast, recommended |

| Q5_K_M | 134.1 GB | |

| Q6_K | 167.6 GB | very good quality |

| Q8_0 | 188.2 GB | fast, best quality |

An MTP (multi-token prediction) draft head is also available for speculative decoding: mtp-cyberneurova-Qwen3.8-Flash-Next-Q8_0.gguf (3.9 GB).

Loading (any current llama.cpp)

llama-server -m cyberneurova-Qwen3.8-Flash-Next.Q6_K.gguf -ngl 99 -fa on -c 32768 -ctk q8_0 -ctv q8_0 --jinja