Provider Tracking
Track AI model fingerprints over time.
Fingerprint History
Loading history data...
Please select a model
Model Cards
Track updates and changes to model and system cards.
Model Card History
Loading model card data...
Please select a model
Usage
Getting Started
Learn how to use stampr_ai to verify AI model identities in your applications. This guide covers installation, basic usage, and advanced features.
Current status: The package is available for experimentation and integration work, but fingerprinting algorithms and verification coverage are still being refined.
Installation
Install the stampr_ai package using pip:
pip install stampr-aiSetup
Set the provider credentials you want to use before running verification:
import os
os.environ["OPENAI_API_KEY"] = "your_openai_api_key_here"
os.environ["OPENROUTER_API_KEY"] = "your_openrouter_api_key_here"
os.environ["HF_TOKEN"] = "your_huggingface_token_here"Python Usage
OpenAI Example
Verify against the latest published fingerprint:
from stampr_ai import get_api_key, verify_tag
import os
api_key = get_api_key("OpenAI") or os.environ.get("OPENAI_API_KEY")
result = verify_tag("gpt-4o", "latest", api_key)
results = result.get("results", {})
if results and all(v.get("verified") for v in results.values()):
print("Fingerprint verified.")
else:
print(f"Verification failed: {result.get('error', 'no match')}")
print(f"Mode: {result.get('mode')}")OpenRouter Example
Verify against a specific published fingerprint hash prefix:
from stampr_ai import get_api_key, verify_tag
import os
api_key = get_api_key("OpenRouter") or os.environ.get("OPENROUTER_API_KEY")
result = verify_tag("Llama4_17b", "bede20", api_key)
results = result.get("results", {})
if results and all(v.get("verified") for v in results.values()):
print("Fingerprint verified.")
else:
print(f"Verification failed: {result.get('error', 'no match')}")
print(f"Mode: {result.get('mode')}")Generate A Reference Fingerprint
Generate a reference fingerprint JSON from a config file:
stampr generate --config config.yaml --api-key OPENAI_API_KEY
# Optional output directory
stampr generate --config config.yaml --api-key OPENAI_API_KEY --out _fingerprints/Command Line Usage
Use the default verify command or the explicit subcommands directly from the shell:
Quick Verify
# Same as: stampr verify gpt-4o --target latest --api-key OPENAI_API_KEY
stampr gpt-4o --api-key OPENAI_API_KEY
stampr verify gpt-4o --target latest --api-key OPENAI_API_KEY
stampr verify gpt-4o --target bede20 --api-key OPENAI_API_KEYTargets And Tracing
# Local fingerprint file
stampr verify gpt-4o --target _fingerprints/<hash>_<timestamp>.json --api-key OPENAI_API_KEY
# URL fingerprint file
stampr verify gpt-4o --target https://example.com/fp.json --api-key OPENAI_API_KEY
# OpenAI-compatible base URL override
stampr verify gpt-4o --target latest --api-key OPENAI_API_KEY --base-url https://my.openai.compat/api
# Structured traces
stampr --trace --trace-dir _traces verify gpt-4o --target latest --api-key OPENAI_API_KEY
--api-key accepts either a literal key or an environment variable name such as OPENAI_API_KEY.
Understanding Results
Verification responses currently expose a few key fields worth checking first:
- results: Per-check verification output returned by the current verification flow
- verified: Nested boolean flags that indicate whether each verification step passed
- mode: The verification mode used, such as latest or specific
- error: Helpful mismatch information when no verification path succeeds
About stampr_ai
What is stampr_ai?
stampr_ai tracks and reports on AI model fingerprint and model/system card changes. The public package currently exposes the verification APIs and CLI so developers can inspect the workflow and experiment with published fingerprints.
The tool consists of two main components: a web-based tracker for monitoring model fingerprints over time, and a Python package for verification workflows in your own applications. The package is currently in alpha, and the fingerprinting algorithms and verification coverage are still evolving.
How It Works
- Reference Fingerprints: Configured collection workflows generate fingerprints and related artifacts for supported models
- Public Verification: The Python package compares runtime model behaviour against published, local, or remote fingerprint targets
- Change Tracking: The website tracks model and card changes over time so users can inspect what has changed and when
Components
Web Tracker
This tracker page provides:
- Visualization of current model fingerprints
- Historical tracking of model changes
- Status context for what is live today versus still in progress
Python Package
The pip package allows you to:
- Run the public verification API and CLI from PyPI
- Integrate verification experiments into your own workflows
- Target the latest, a specific hash prefix, or a local or remote fingerprint file
pip install stampr-aiContact
Questions? Collaboration ideas? We'd love to hear from you.