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.

🚀 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

⚡ 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

📄 License

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