---
id: openinference-semantic-conventions
version: "0.1.32"
license: Apache-2.0
license_treatment: permissive
maintenance: active
---

# openinference-semantic-conventions — OpenInference Semantic Conventions

License: permissive · Maintenance: active · Downloads: 8.5M/mo

## What it is and what it does

OpenInference Semantic Conventions is a specification library that standardizes how AI application traces—particularly LLM calls, vector store retrievals, and external tool invocations—are represented and labeled. It works alongside OpenTelemetry to provide a common vocabulary for observability across AI systems, enabling consistent tracing whether you use Arize Phoenix, Arize AX, or any other OpenTelemetry-compatible backend.

The package itself is a pure Python module with no external runtime dependencies, making it lightweight to integrate into instrumentation libraries and AI frameworks. It defines the attribute names, semantic meanings, and conventions that other OpenInference instrumentation packages (for LangChain, LlamaIndex, OpenAI, and others) use when emitting trace data. If you are building or using an AI application that needs observability, this package provides the foundational schema that ensures your traces are interpreted consistently across tools.

Use it for:

- Standardize trace attributes when instrumenting LLM applications with OpenTelemetry for observability.
- Ensure consistent semantic meaning of trace events across different AI frameworks and backends.
- Build custom instrumentation libraries that emit traces following OpenInference conventions.
- Integrate with Arize Phoenix or other OpenTelemetry backends using a common trace schema.
- Document and enforce tracing standards for retrieval-augmented generation (RAG) and agent workflows.

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

Defines semantic conventions and standardized attributes for tracing AI applications, particularly LLM invocations, retrieval operations, and tool usage, for use with OpenTelemetry-compatible backends.

Yes, if you are building or instrumenting AI applications and need a shared semantic vocabulary for tracing. It is lightweight, actively maintained, and essential infrastructure for any project using OpenInference instrumentation libraries or OpenTelemetry with AI workloads. No security vulnerabilities and permissive licensing. Install it as a dependency of instrumentation packages rather than directly unless you are defining custom tracing logic.

## Install

pip install openinference-semantic-conventions
uv add openinference-semantic-conventions
poetry add openinference-semantic-conventions

## Installing openinference-semantic-conventions

Before you install: Low friction; pure Python wheel with no runtime dependencies. Actively maintained with a release 7 days ago and 1147 repository stars.

License in practice: Apache-2.0 permissive license allows commercial and private use with minimal restrictions.

Quickstart:

pip install openinference-semantic-conventions

from openinference_semantic_conventions import ...
# Import semantic convention constants for use in tracing instrumentation

Requires Python 3.10 or later (supports up to 3.14).

Verify before relying:

- Whether this package alone provides tracing without a separate instrumentation library or OpenTelemetry backend.
- What specific semantic convention constants or classes are exported and how they are typically used in practice.

## Package facts

- License: Apache-2.0 (permissive)
- Python support: supports_current
- Install friction: low
- Maintenance: active
- Downloads: 8.5M/month (top 5,000 on PyPI)
- Known vulnerabilities: none known

## Tags

llm tracing conventions, opentelemetry semantic standards, ai application instrumentation, llm observability schema, tracing attributes for ai apps, openinference specifications, vector store retrieval tracing, llm-observability, opentelemetry, ai-tracing

[View on SkillFed](https://skillfed.io/packages/openinference-semantic-conventions) · [View on PyPI](https://pypi.org/project/openinference-semantic-conventions/)