inspect-ai
Framework for large language model evaluations
Decision gist · record as of 2026-08-14
Yes, if you need to evaluate LLMs systematically. The framework is actively maintained, permissively licensed, and backed by a government security institute. The 40 runtime dependencies and Python 3.10+ requirement are substantial but typical for a full-featured evaluation platform. Install if you're building evaluation pipelines or comparing model outputs; skip if you only need lightweight model testing.AI-flagged interpretation of the facts on this page — verify before relying
Before you install
- Requires Python 3.10 or later.
- 40 runtime dependencies including pydantic, fastapi, boto3, and numpy—ensure your environment can resolve them.
- Low friction installation with a wheel distribution.
License · maintenance · safety
MIT License (permissive) — MIT License permits commercial and private use with minimal restrictions, making it suitable for most projects without licensing concerns.
last release 2026-08-12 (2 days) · last repo commit 2026-08-14 · 2,549 stars
0 known vulnerabilities (OSV.dev, 2026-08-14) · 12,105,306 downloads/mo, #1,342 on PyPI
Alternatives
Verify before relying
pip install inspect-ai
import inspect_ai
# See https://inspect.aisi.org.uk/ for evaluation setup and model configuration- Whether the 200+ pre-built evaluations are sufficient for your specific model types and use cases
- Performance characteristics when running evaluations against multiple models concurrently
- Extensibility mechanism details for custom elicitation and scoring techniques beyond built-in components
What it is and what it does
Inspect is a framework for evaluating large language models, created by the UK AI Security Institute. It provides a structured environment for running evaluations on any model, with built-in support for prompt engineering, tool use, multi-turn dialogue, and model-graded scoring. The framework includes over 200 pre-built evaluations ready to run out of the box.
The package has a substantial dependency graph (40 runtime dependencies including pydantic, fastapi, boto3, numpy, and AWS integration libraries) reflecting its role as a full-featured evaluation platform. It supports both Python development workflows and web-based UI interaction through a TypeScript/React frontend. Active maintenance and recent releases indicate ongoing development.
Use it for
- Run standardized evaluations against multiple LLM providers to compare model performance on safety, reasoning, and capability benchmarks
- Build custom evaluation pipelines combining prompt engineering, tool use, and model-graded scoring for domain-specific tasks
- Integrate LLM evaluation into CI/CD workflows to validate model behavior before deployment
- Develop and share new evaluation techniques as Python extensions compatible with the Inspect framework
Worth the install?
AI-flagged interpretation of the facts on this page. Verify before relying on it.
Yes, if you need to evaluate LLMs systematically.
The framework is actively maintained, permissively licensed, and backed by a government security institute. The 40 runtime dependencies and Python 3.10+ requirement are substantial but typical for a full-featured evaluation platform. Install if you're building evaluation pipelines or comparing model outputs; skip if you only need lightweight model testing.
Install
inspect-ai on PyPI
Before you install
Low friction installation with a wheel distribution. Active maintenance—released 2 days ago with a commit history through 2026-08-14. Requires Python 3.10 or later.
Requires Python 3.10 or later. 40 runtime dependencies including pydantic, fastapi, boto3, and numpy—ensure your environment can resolve them.
License in practice
MIT License permits commercial and private use with minimal restrictions, making it suitable for most projects without licensing concerns.
Quickstart
pip install inspect-ai
import inspect_ai
# See https://inspect.aisi.org.uk/ for evaluation setup and model configurationVerify before relying
- Whether the 200+ pre-built evaluations are sufficient for your specific model types and use cases
- Performance characteristics when running evaluations against multiple models concurrently
- Extensibility mechanism details for custom elicitation and scoring techniques beyond built-in components
Package facts
| License | MIT License permissive |
| Python support | Supports the current Python release >=3.10 |
| Install friction | Low. Pure-Python wheel |
| Runtime dependencies | 40 packagesagent-client-protocolaioboto3anyiobeautifulsoup4boto3clickdebugpydocstring-parserexceptiongroupfastapifsspechttpxijsonjsonlinesjsonpatchjsonpath-ngjsonrefjsonschemammh3nest_asyncio2numpyplatformdirspsutilpydanticpython-dotenvpyyamlrichs3fssemvershortuuid |
| Maintenance | Actively maintained 2 days since the last release |
| Last repo commit | |
| First released | |
| Downloads | 12,105,306 / month, #1,342 on PyPI 30-day window, as of 2026-08-14 |
| Known vulnerabilities | None known OSV.dev, checked 2026-08-14 |
| Classifiers | Development Status :: 4 - BetaEnvironment :: ConsoleIntended Audience :: DevelopersIntended Audience :: Science/ResearchLicense :: OSI Approved :: MIT LicenseNatural Language :: EnglishOperating System :: OS IndependentProgramming Language :: Python :: 3Topic :: Scientific/Engineering :: Artificial IntelligenceTyping :: Typed |
Evidence: inspect_ai-0.3.258-py3-none-any.whl
Tags
Let your AI agent find packages like this
Example. Real query, live index.
You found this page by searching. An agent finds it by wishing: SkillFed indexes 14,416 PyPI packages by what they can do, searchable in plain language.
wish › “model grading system”
- inspect-aiInspect is a framework for evaluating large language models,…
- autoevalsAutoevals provides automatic evaluation methods for AI model outputs,…
- rubricRubric is a Python library for evaluating LLM outputs against…
Give your agent the search over MCP, or paste the wish link into any chat.
More Artificial Intelligence packages
LiteLLM provides a unified Python interface to call 100+ LLM providers (OpenAI, Anthropic, Gemini, Bedrock, Azure, and others) using OpenAI-compatible API format, available as both a Python SDK and a self-hosted AI Gateway proxy server.
Install it if you need to work with multiple LLM providers or want to centralize LLM routing in your organization.
Client library and CLI tool for downloading, uploading, and managing models, datasets, and repositories on the Hugging Face Hub platform.
Install it if you work with Hugging Face Hub models or datasets.
LangChain provides a framework for building agents and LLM-powered applications by composing language models, tools, and memory through a unified API that abstracts over multiple model providers.
hf-xet provides chunk-based deduplication and efficient file transfer for the Hugging Face Hub, enabling faster uploads and downloads of large files with local disk caching.
Tokenizers converts raw text into token sequences for NLP models, with support for training custom vocabularies and using pre-built tokenizers (BPE, WordPiece) optimized for speed via Rust.
Transformers provides a unified framework for loading, fine-tuning, and running state-of-the-art pretrained models across text, vision, audio, video, and multimodal tasks using PyTorch, JAX, or TensorFlow.
Install it if you need to run or train any transformer-based model for NLP, vision, audio, or multimodal tasks.
See also inspect-evals · inspect-scout · inspect-swe · autoevals · vellum-ai · nemo-evaluator · deepeval · strands-agents-evals · bfcl-eval · deepteam