---

license: apache-2.0

base_model:

- Qwen/Qwen3.5-9B

pipeline_tag: text-generation

library_name: transformers

tags:

- 1m-context

- uncensored

- reasoning

- long-context

- coding

- gguf

- llama-cpp

---

!PINQWEN-3.5-9B-1M

<p align="center">

<img src="https://img.shields.io/badge/context-1%2C000%2C000_tokens-8A2BE2?style=for-the-badge" alt="1M context">

<img src="https://img.shields.io/badge/alignment-UNCENSORED-c0143c?style=for-the-badge" alt="Uncensored">

<img src="https://img.shields.io/badge/format-GGUF-1f6feb?style=for-the-badge" alt="GGUF">

<img src="https://img.shields.io/badge/decode-MTP_speculative-ff8c00?style=for-the-badge" alt="MTP">

<img src="https://img.shields.io/badge/license-Apache_2.0-2ea44f?style=for-the-badge" alt="Apache 2.0">

</p>

PINQWEN‑3.5‑9B‑1M — GGUF

> ### 🪟 *A 9B that reads a million tokens, thinks before it speaks, and never refuses — run it on your own machine.*

PINQWEN‑3.5‑9B‑1M is Blackfrost AI's compact powerhouse — a reasoning model distilled through The Void, our multi‑teacher reasoning‑distillation method, on the Qwen 3.5 9B architecture. Reasons first, answers second (<think> → answer), across a full one‑million‑token context window. This is the complete GGUF drop for llama.cpp — every quant, the vision projector, and the MTP speculative‑decode head for faster generation.

<!-- 👉 hero image goes here -->

---

🚀 Why PINQWEN‑3.5‑9B‑1M?

• 🪟 1,000,000‑token context — whole codebases and books in one prompt, YaRN‑extended for efficient long context.
• 🔓 Uncensored — answers directly, no reflexive refusals.
• 🧠 Reasons in the open — native <think>…</think> chain‑of‑thought from a panel of frontier teachers.
• ⚡ MTP speculative decode — ships with the multi‑token‑prediction head for accelerated generation.
• 👁 Vision projector included (mmproj-*) for image input in compatible runtimes.

📦 This repository — GGUF: pick your quant

| File | Precision | Size | Notes |

|---|---|---|---|

| PINQWEN-3.5-9B-Q2_K.gguf | 2‑bit | ~3.7 GB | Smallest; tight budgets |

| PINQWEN-3.5-9B-Q3_K_M.gguf| 3‑bit | ~4.5 GB | |

| PINQWEN-3.5-9B-Q4_K_M.gguf| 4‑bit | ~5.4 GB | ⭐ Recommended balance |

| PINQWEN-3.5-9B-Q5_K_M.gguf| 5‑bit | ~6.2 GB | Higher quality |

| PINQWEN-3.5-9B-Q6_K.gguf | 6‑bit | ~7.1 GB | |

| PINQWEN-3.5-9B-Q8_0.gguf | 8‑bit | ~9.2 GB | Near‑lossless |

| PINQWEN-3.5-9B-f16.gguf | f16 | ~18 GB | Full precision |

| mmproj-PINQWEN-3.5-9B-BF16.gguf | — | ~0.9 GB | Vision projector (image input) |

⚡ Quickstart (llama.cpp)

# text
llama-cli -m PINQWEN-3.5-9B-Q4_K_M.gguf -p "Explain the CAP theorem." -ngl 99

# thinking off (direct answers)
llama-server -m PINQWEN-3.5-9B-Q4_K_M.gguf -ngl 99 \
  --chat-template-kwargs '{"enable_thinking":false}'

# with vision
llama-mtmd-cli -m PINQWEN-3.5-9B-Q4_K_M.gguf \
  --mmproj mmproj-PINQWEN-3.5-9B-BF16.gguf --image pic.jpg -p "Describe this."

🎛 Formats — every one is 1M context

| Format | Repository | Best for |

|---|---|---|

| BF16 | PINQWEN-3.5-9B-1M-BF16 | Reference precision, fine‑tuning, vision |

| NVFP4 | PINQWEN-3.5-9B-1M-NVFP4 | Fast serving on NVIDIA Blackwell |

| GGUF | PINQWEN-3.5-9B-1M-GGUF | llama.cpp / local — full quant ladder + MTP |

📄 License

Apache‑2.0. Base architecture: Qwen 3.5 9B. As an uncensored model, use responsibly.

---

<p align="center"><em>PINQWEN‑3.5‑9B‑1M — part of Blackfrost AI's <b>Void</b> model family. 🖤</em></p>