# Opik review

> Opik is an open-source platform from Comet for tracing, evaluating, and monitoring LLM applications, RAG systems, and AI agents.

Devtoolwatch Board, the week of Oct 25–31, 2026. Canonical page: https://index.devtoolwatch.com/tools/opik/

## TL;DR

Opik is an Apache-2.0 platform for LLM observability, evaluation, prompt management, and production monitoring, built by Comet. It is for ML engineers and AI teams moving LLM apps and agents from prototype to production. The full platform, backend included, can be self-hosted, and Opik is framework-agnostic.

## Opik at a glance

- **Category:** [Evals and observability](https://index.devtoolwatch.com/categories/evals-and-observability/)
- **Maintained by:** Comet (organization, [@comet-ml](https://github.com/comet-ml) on GitHub)
- **License:** Apache 2.0
- **Language:** Python, TypeScript, Shell
- **Repository:** [github.com/comet-ml/opik](https://github.com/comet-ml/opik)
- **Website:** [www.comet.com/docs/opik](https://www.comet.com/docs/opik/)
- **npm package:** [opik](https://www.npmjs.com/package/opik) 2.2.113
- **PyPI package:** [opik](https://pypi.org/project/opik/) 2.2.113
- **Latest release:** [2.2.113](https://github.com/comet-ml/opik/releases/tag/2.2.113) on GitHub, Oct 21, 2026
- **First published:** May 10, 2023

## What Opik does

- Opik traces LLM calls, tool executions, retrieval steps, and agent activity, with full trace trees for multi-step agents.
- Opik runs evaluations with datasets, experiments, code-based metrics, and LLM-as-a-judge metrics for hallucination, moderation, and RAG assessment.
- Opik monitors feedback scores, trace counts, and token usage in dashboards, and scores incoming traces with online evaluation rules.
- Opik stores and versions prompts and offers a Prompt Playground for experimenting with prompts and models.
- Opik provides a PyTest integration to evaluate LLM pipelines in CI/CD on every commit.
- Opik includes an MCP server so Claude Code, Cursor, VS Code Copilot, Codex, or opencode can read traces and run evaluations from chat.

### How it works

Opik runs as a server with a web app, either on Comet.com Cloud or self-hosted with Docker Compose or Kubernetes and Helm. Your application sends traces through the Python or TypeScript SDK, a REST API, or OpenTelemetry. In Python, you wrap functions with the @track decorator to log calls, including nested calls. Integrations cover frameworks such as Google ADK, Autogen, and Flowise AI.

## How to install Opik

**Python SDK with pip**

```
pip install opik
```

**Connect a coding agent via MCP**

```
uvx opik mcp configure
```

**Self-host on Linux or Mac (from the cloned repo)**

```
./opik.sh
```

### Requirements

- The Python SDK requires Python 3.10 or later.
- The TypeScript and JavaScript SDK requires Node.js 18 or later.
- The MCP setup command needs uv.
- Either a Comet account for Opik Cloud, or Docker or Kubernetes to self-host.

### First steps

1. Install the Python SDK, then run opik configure to point it at Opik Cloud or your self-hosted server.
2. Wrap a function with the @track decorator so every call, including nested calls, is logged to Opik.
3. For a local self-hosted instance, run ./opik.sh in the cloned repo and open localhost:5173 in your browser.
4. Turn a wrong-looking trace into a test case through the UI or SDK, then run the test suite to verify fixes.

## Who Opik is for

- ML engineers building LLM-powered agents who need trace trees for multi-step agents and tool calls.
- AI teams moving LLM apps from prototype to production monitoring.
- Teams that need self-hosted observability with no data leaving their own environment.
- Developers who want LLM evaluations to run in CI through PyTest.

### Limitations

- Opik's Docker Compose setup is aimed at local development and testing. Production self-hosting uses Kubernetes and Helm.
- Opik's one-click MCP badges and add-mcp fallback target Opik Cloud only. Self-hosted setups use the uvx command.
- Opik's native SDKs cover Python and TypeScript. Other languages such as Java and Ruby connect through OpenTelemetry.
- Running ./opik.sh --clean removes all Opik data volumes, and all Opik data is lost.

## Usage data (the week of Oct 25–31, 2026)

- Board rank: #75 overall, #5 in Evals and observability, momentum score 69
- Downloads that week: 709.1k (+33% vs 4-week average)
- New GitHub stars per week (last 12 weeks): Aug 9: 190; Aug 16: 143; Aug 23: 146; Aug 30: 158; Sep 6: 159; Sep 13: 187; Sep 20: 111; Sep 27: 126; Oct 4: 149; Oct 11: 215; Oct 18: 207; Oct 25: 190

## Opik alternatives

| Rank | Tool | Stars | Downloads/wk |
|---|---|---|---|
| #1 | [SkillSpector](https://index.devtoolwatch.com/tools/nvidia-skillspector/) | 23.7k | – |
| #2 | [Judgeval](https://index.devtoolwatch.com/tools/judgmentlabs-judgeval/) | 1.1k | 80.2k |
| #3 | [A.I.G (AI-Infra-Guard)](https://index.devtoolwatch.com/tools/tencent-ai-infra-guard/) | 7.2k | – |
| #4 | [Langfuse](https://index.devtoolwatch.com/tools/langfuse/) | 36.8k | 9.8M |
| #6 | [promptfoo](https://index.devtoolwatch.com/tools/promptfoo/) | 26.7k | 1.2M |
| #7 | [garak](https://index.devtoolwatch.com/tools/garak/) | 9.8k | 12.9k |

## Opik FAQ

### Is Opik free and open source?

Yes. Opik's source code is on GitHub (comet-ml/opik) under the Apache 2.0 license, so you can use it for free. Its website also links to a pricing page (https://www.comet.com/site/pricing/) for any paid plans.

### How do I install Opik?

Run `pip install opik`. It needs Node.js >=18. The install section above has the other options from the project's README.

### Is Opik still maintained?

Yes. The comet-ml/opik repository was last updated on Oct 22, 2026, and the latest release, 2.2.113 on GitHub, came out on Oct 21, 2026.

### How popular is Opik?

Opik has 23.1k GitHub stars as of Nov 1, 2026. It gained 190 new stars in the week of Oct 25–31, 2026, against a 12-week average of 165 a week. Its main package had about 709.1k downloads that week, +33% against its own 4-week average. That made it #5 among LLM evals and observability tools on the Devtoolwatch Board.

### Opik vs Langfuse: how do they compare?

Opik and Langfuse are both tracked by Devtoolwatch as LLM evals and observability tools. Opik has 23.1k GitHub stars and Langfuse has 36.8k. In the week of Oct 25–31, 2026, Opik's main package had 709.1k downloads and Langfuse's had 9.8M. Langfuse is an open source LLM engineering platform for tracing, evaluating, prompt management, and debugging LLM applications and agents. [Langfuse review](https://index.devtoolwatch.com/tools/langfuse/)

### Opik vs MLflow: how do they compare?

Opik and MLflow are both tracked by Devtoolwatch as LLM evals and observability tools. Opik has 23.1k GitHub stars and MLflow has 28.6k. In the week of Oct 25–31, 2026, Opik's main package had 709.1k downloads and MLflow's had 5.2M. MLflow is an open source AI engineering platform for tracing, evaluating, monitoring, and managing agents, LLM applications, and ML models. [MLflow review](https://index.devtoolwatch.com/tools/mlflow/)

### What are the best alternatives to Opik?

Other LLM evals and observability tools on the Devtoolwatch Board in the week of Oct 25–31, 2026, by momentum: SkillSpector (#1), Judgeval (#2), A.I.G (AI-Infra-Guard) (#3), Langfuse (#4), promptfoo (#6). The alternatives table above has their numbers.

### Is Opik open source?

Yes. Opik is licensed under Apache 2.0. The server, web application, and core observability and evaluation features can be self-hosted without a commercial license.

### Can I self-host Opik?

Yes. Opik can run locally with Docker Compose using the opik.sh or opik.ps1 script, or on Kubernetes with a Helm chart for larger deployments.

### Does Opik support AI agent tracing?

Yes. Opik captures multi-step traces containing LLM calls, tool executions, retrieval steps, and other agent activity.

### Is Opik tied to a specific agent framework?

No. Opik is framework-agnostic. It supports its own SDK, OpenTelemetry, and framework-specific integrations, and works with OpenAI, Anthropic, LangChain, and many other providers and frameworks.

### What evaluation metrics does Opik include?

Opik offers LLM-as-a-judge and heuristic metrics, including hallucination detection, moderation, answer relevance, context precision, and context recall.

## Sources

1. [comet-ml/opik on GitHub](https://github.com/comet-ml/opik)
2. [Opik README](https://github.com/comet-ml/opik/blob/main/README.md)
3. [opik on npm](https://www.npmjs.com/package/opik)
4. [opik on PyPI](https://pypi.org/project/opik/)
5. [Opik website](https://www.comet.com/docs/opik/)
6. [Release 2.2.113 (Oct 21, 2026)](https://github.com/comet-ml/opik/releases/tag/2.2.113)
7. [Release 2.2.112 (Oct 21, 2026)](https://github.com/comet-ml/opik/releases/tag/2.2.112)
8. [Release 2.2.111 (Oct 21, 2026)](https://github.com/comet-ml/opik/releases/tag/2.2.111)
9. [Hacker News: Show HN: Opik, an open source LLM evaluation framework](https://news.ycombinator.com/item?id=41567192)
10. [Hacker News: Open-Source SDK for Automated Prompt and Agent Optimization](https://news.ycombinator.com/item?id=44062034)
11. [Hacker News: Opik Agent Optimizer – Open-Source Prompt Optimization Framework](https://news.ycombinator.com/item?id=46267143)
12. [Devtoolwatch Board, the week of Oct 25–31, 2026](https://index.devtoolwatch.com/)