🔬 probe
Behavioral testing framework for LLMs — test properties, not benchmarks.
Benchmarks tell you a model scores 85% on MMLU.
probe tells you whether the model actually behaves consistently on your tasks.
pip install -e ".[openai]"
Quick Start
Python
from probe import quick_test, print_summary
results = quick_test(
model="openai:gpt-4o-mini",
inputs=[
"What is the capital of France?",
"Is 17 a prime number?",
"Explain photosynthesis in 2 sentences.",
],
properties=["consistency", "robustness"],
)
print_summary(results)
CLI
probe run --model openai:gpt-4o-mini --input "What is the capital of France?" -p consistency
probe run --model openai:gpt-4o-mini --inputs test_cases.txt -p consistency,invariance,robustness
probe compare --model-a openai:gpt-4o --model-b ollama:llama3 --inputs test_cases.txt
probe list-properties
Behavioral Properties
| Property | What It Tests |
|---|---|
| consistency | Rephrase question N ways → same answer? |
| invariance | Change irrelevant details → answer holds? |
| negation | Negate the question → answer flips? |
| robustness | Add typos → model still works? |
Providers
openai:gpt-4o-mini # OpenAI
anthropic:claude-sonnet-4-20250514 # Anthropic
ollama:llama3 # Local via Ollama
License
MIT
1 maintainer
Metadata
1 maintainer
Release files for probe-llm 0.1.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| probe_llm-0.1.0.tar.gz | 12.8 kB | Without provenance | Details for probe_llm-0.1.0.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| probe_llm-0.1.0-py3-none-any.whl | Python 3 | none | any | Details for probe_llm-0.1.0-py3-none-any.whl |
Total release size: 31.7 kB
Release files / probe_llm-0.1.0.tar.gz
| Download URL | probe_llm-0.1.0.tar.gz |
|---|---|
| Size | 12.8 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
3d7330a02e9827d116a4730971cc011a3658d00816da1af4a069be575f9e4a2e
|
|
BLAKE2b-256 checksum How to use checksums |
cc32b30871a24709f24287bcc4d8518f20a3f74cee9bd85f262b36994b5f4601
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.12.3
|
Release files / probe_llm-0.1.0-py3-none-any.whl
| Download URL | probe_llm-0.1.0-py3-none-any.whl |
|---|---|
| Size | 18.8 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
3d6599b6d14af8b704312f32e29ef4f28317fed042d23efee622bb646fba627b
|
|
BLAKE2b-256 checksum How to use checksums |
fb71feae4a835e4af7830222c3b84c4aedbcb8579c082550b7ddc3abfaa264e5
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.12.3
|
Security for probe-llm 0.1.0
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.
