Introducing Pirate Face
The first verifiable torrent index for AI - every Hugging Face open model, provably identical to the original and kept alive forever by the swarm.
https://t.co/6UpdzEuYTI
The permanence layer for AI. How it works π
Community activity Β· 24 Sept 2026
queencaffeine verified their Hugging Face identity on Pirate Face β
https://t.co/JF4EaEIe5w
More in this thread β
By popular demand, Models > filter by Parameters and Activated Parameters is now available.
e.g. a search for '200Bβ300B total' and '20Bβ30B activated' returns Qwen/Qwen3-235B-A22B and OpenGVLab/InternVL3_5-241B-A28B.
Models page β
Pirate Face power user @_QCAI is quickly rising up the leaderboard.
She already rescued and is seeding 10 models on Pirate Face, all hash-verified against Hugging Face:
DeepSeek-V4-Flash (GGUF)
olmOCR-2-7B (GGUF)
Qwen3-VL-2B-Instruct (GGUF)
Qwen3-Embedding-0.6B (GGUF)
Qwen2.5-Coder-1.5B-Instruct (GGUF)
Qwen3-1.7B (GGUF)
PaddleOCR-VL-1.6
GLM-OCR
whisper.cpp (large-v3-turbo)
Kokoro-82M
I just seeded DeepSeek-V4-Flash on Pirate Face so it stays available even if it disappears from its original hosts.
Actively seeding it right now. No company and no country should be able to strip people's access to free, open-source models.
Link below!
Here's your invite to come help me keep this alive or download whatever models you want to preserve. All have matching hashes with HuggingFace so you know you're getting the real thing that hasn't been tampered with.
https://t.co/XXomBc3tqh