Galileo

aid: galileo-ai
url: https://raw.githubusercontent.com/api-evangelist/galileo-ai/refs/heads/main/apis.yml
name: Galileo
x-type: company
description: >-
  Galileo (galileo.ai, formerly rungalileo.io) is the GenAI evaluation, observability, and production guardrail platform
  for LLM and agentic applications. Galileo provides pre-built and custom evaluators, agentic trace and span logging,
  dataset and experiment management, prompt management, runtime protect (guardrails), and integrations with major LLM
  providers and agent frameworks. The platform exposes a public REST API plus official Python and TypeScript SDKs and
  integrates with LangChain, CrewAI, OpenAI Agents, and OpenTelemetry/OpenInference. Galileo is deployable as SaaS, VPC,
  or on-prem.
image: https://kinlane-images.s3.amazonaws.com/shared/apis-json/apis-json-logo.jpg
tags:
  - AI Evaluation
  - AI Observability
  - GenAI
  - Guardrails
  - Agentic AI
  - LLM
  - Tracing
  - Experiments
  - Prompts
  - Datasets
created: '2026-05-08'
modified: '2026-05-25'
specificationVersion: '0.20'
apis:
  - aid: galileo-ai:platform
    name: Galileo Public API
    description: >-
      Public REST API for the Galileo platform (api.galileo.ai/v2). Provides projects, datasets, experiments, traces and
      spans, log streams, prompts, annotations, groups, integrations, users, API keys, and authentication.
      Authentication is via a Galileo-API-Key header or via OAuth2-password bearer token obtained at /v2/token. The
      canonical OpenAPI is published at https://api.galileo.ai/public/v2/openapi.json and a documentation index is
      available at https://docs.galileo.ai/llms.txt.
    humanURL: https://docs.galileo.ai/
    baseURL: https://api.galileo.ai
    tags:
      - Projects
      - Datasets
      - Experiments
      - Traces
      - Spans
      - LogStreams
      - Prompts
      - Annotations
      - Groups
      - Integrations
      - ApiKeys
      - Users
      - Auth
    properties:
      - type: Documentation
        url: https://docs.galileo.ai/
      - type: APIReference
        url: https://docs.galileo.ai/api-reference/
      - type: OpenAPI
        url: openapi/galileo-ai-openapi.yml
      - type: CanonicalOpenAPI
        url: https://api.galileo.ai/public/v2/openapi.json
      - type: LLMsTxt
        url: https://docs.galileo.ai/llms.txt
      - type: GettingStarted
        url: https://docs.galileo.ai/getting-started/quickstart
      - type: Quickstart
        url: https://docs.galileo.ai/getting-started/quickstart
      - type: PythonSDK
        url: https://pypi.org/project/galileo/
      - type: PythonSDKRepo
        url: https://github.com/rungalileo/galileo-python
      - type: PythonSDKReference
        url: https://docs.galileo.ai/sdk-api/python/sdk-reference
      - type: TypeScriptSDK
        url: https://www.npmjs.com/package/galileo
      - type: TypeScriptSDKRepo
        url: https://github.com/rungalileo/galileo-js
      - type: TypeScriptSDKReference
        url: https://docs.galileo.ai/sdk-api/typescript/sdk-reference
      - type: Authentication
        url: https://docs.galileo.ai/api-reference/auth/
      - type: Concepts
        url: https://docs.galileo.ai/concepts/
      - type: Metrics
        url: https://docs.galileo.ai/concepts/metrics/overview
      - type: Logging
        url: https://docs.galileo.ai/concepts/logging/overview
      - type: Protect
        url: https://docs.galileo.ai/concepts/protect/overview
      - type: SamplesRepo
        url: https://github.com/rungalileo/sdk-examples
      - type: Vocabulary
        url: vocabulary/galileo-ai-vocabulary.yml
      - type: JSONLDContext
        url: json-ld/galileo-ai-context.jsonld
      - type: SpectralRules
        url: rules/galileo-ai-rules.yml
    features:
      - name: Evaluations
        description: >-
          20+ pre-built evaluators for RAG, agents, safety, and security plus custom evaluators built in the console or
          via API.
      - name: Luna Models
        description: Lightweight in-house judge models that replace expensive LLM-as-judge calls at substantially lower cost.
      - name: Tracing & Spans
        description: >-
          Capture nested traces (workflow, LLM, retriever, tool) for GenAI and agentic apps via SDK or
          OTel/OpenInference.
      - name: Experiments
        description: Run prompt or runner experiments against datasets with attached metrics; compare across runs.
      - name: Datasets
        description: Manage datasets with rows, columns, ground-truth, and synthetic data; query and download content.
      - name: Prompts
        description: Versioned prompt templates with parameters used in experiments and production.
      - name: Annotations
        description: Templates and ratings for capturing human/SME feedback on spans and outputs.
      - name: Protect (Guardrails)
        description: Runtime policy enforcement converting eval metrics into guardrails that block or rewrite harmful responses.
      - name: Log Streams
        description: Production log streams group traces by environment for monitoring and analysis.
      - name: Integrations
        description: Available integrations for LLM providers, vector stores, and frameworks managed via /v2/integrations.
      - name: Groups & Access Control
        description: Group roles, group memberships, and per-project access controls.
      - name: API Keys
        description: Per-user API keys, listable, createable, and deletable via the API.
      - name: Multi-Deployment
        description: SaaS, Virtual Private Cloud (VPC), and on-prem deployment options.
    useCases:
      - name: Evaluate RAG Pipelines
        description: Score retrieval relevance, groundedness, and answer correctness across dataset rows.
      - name: Monitor Agentic Workflows
        description: Trace multi-step agent runs, identify failure modes, and convert insights into guardrails.
      - name: Pre-Production Regression Testing
        description: Run experiments against benchmark datasets when shipping a new prompt, model, or chain.
      - name: Runtime Safety
        description: Block or rewrite harmful, off-policy, or hallucinated outputs in production using Protect.
      - name: Prompt A/B Testing
        description: Compare prompt templates or models on the same dataset with shared metrics.
      - name: Human-in-the-Loop QA
        description: Collect SME ratings via annotation templates to refine evaluators and models.
    integrations:
      - name: OpenAI
        description: Wrap OpenAI clients to auto-instrument calls (wrapOpenAI / OpenAI integration).
      - name: LangChain
        description: GalileoCallback handler for LangChain and LangGraph workflows.
      - name: CrewAI
        description: Third-party integration for CrewAI multi-agent workflows.
      - name: OpenAI Agents SDK
        description: GalileoTracingProcessor for the OpenAI Agents framework.
      - name: OpenTelemetry / OpenInference
        description: OTel-compatible ingestion via OpenInference semantic conventions.
      - name: NVIDIA NeMo Agent Toolkit
        description: Integration patterns demonstrated in rungalileo/NeMo-Agent-Toolkit.
      - name: MCP (Model Context Protocol)
        description: MCP server pattern for exposing Galileo to AI agents (rungalileo/demo-mcp).
    solutions:
      - name: GenAI Reliability
        description: Reduce hallucination and incorrect agent behavior via evaluation-driven development.
      - name: AI Safety & Compliance
        description: Enforce safety, security, and policy via Protect guardrails at runtime.
      - name: Cost Optimization
        description: Use Luna models and sampling to reduce LLM-as-judge cost.
common:
  - type: Website
    url: https://galileo.ai/
  - type: DeveloperPortal
    url: https://docs.galileo.ai/
  - type: Documentation
    url: https://docs.galileo.ai/
  - type: APIReference
    url: https://docs.galileo.ai/api-reference/
  - type: Pricing
    url: https://galileo.ai/pricing
  - type: Blog
    url: https://galileo.ai/blog
  - type: GitHubOrganization
    url: https://github.com/rungalileo
  - type: LinkedIn
    url: https://www.linkedin.com/company/galileo-ai
  - type: Plans
    url: plans/galileo-ai-plans-pricing.yml
  - type: RateLimits
    url: rate-limits/galileo-ai-rate-limits.yml
  - type: FinOps
    url: finops/galileo-ai-finops.yml
  - type: JSONLD
    url: json-ld/galileo-ai-context.jsonld
  - type: Vocabulary
    url: vocabulary/galileo-ai-vocabulary.yml
  - type: SpectralRules
    url: rules/galileo-ai-rules.yml
maintainers:
  - FN: Kin Lane
    email: kin@apievangelist.com