Verify model identity and detect downgrades or swaps through fingerprint comparison.
-
Updated
Mar 17, 2026 - Python
Verify model identity and detect downgrades or swaps through fingerprint comparison.
LLM distillation detection & model fingerprinting via statistical forensics — behavioral probing, stylistic signatures & representation similarity. CLI + MCP ready.
PAMELA 行为指纹识别系统:基于单 token 输出分布的 Jensen-Shannon Divergence 度量,识别 LLM API 真实身份。内置 173 个模型参考指纹库(涵盖 OpenAI/Anthropic/Google/Meta/Qwen/DeepSeek/GLM/Mistral/Moonshot/Llama/Nvidia/Amazon/Nous/Cohere/ByteDance 等 30+ 厂商),工具无关——任何能调 LLM API 的客户端(curl/Python/浏览器/Hermes/OpenClaw/Codex)都能采集。Apache-2.0/MIT/CC-BY-4.0
Verify your Claude Code endpoint really serves GLM-5.2 — tokenizer fingerprint + 1M context probe
Authenticity & behavior probe for Claude / Anthropic-compatible endpoints — 47 probes, sandbox fallback, golden similarity, community probe packs. macOS · Linux · Windows.
Add a description, image, and links to the model-fingerprint topic page so that developers can more easily learn about it.
To associate your repository with the model-fingerprint topic, visit your repo's landing page and select "manage topics."