Opik is an open source AI Observability and Evals platform launched by Comet, covering Trace, Test Suites, Prompt optimization and production alerts, and supports cloud and self-hosted deployment.

Opik

Core parameters and statistics

Projects	Public Information
Product Positioning	Open-source AI Observability, Evaluation, and Optimization
Key Capabilities	Tracing, LLM-as-a-Judge, Test Suites, Prompt/Tool Optimization
Deployment Path	Open Source Self-Hosted + Free/Pro/Enterprise Cloud
Integration scale	Pricing page public 40+ integration ecosystem
Production scale signal	README announces design target 40M+ traces/day
Community size	GitHub 19.4k+ stars, 1.5k forks
Latest version	2.0.58 (2026-06-08)

Product Boundary: Opik focuses on AI application observation, evaluation and optimization, and does not undertake general data warehouse or model training tasks.

User and market recognition

Market Recognition Signal: The official website displays brand logos such as Uber, Netflix, Autodesk, Etsy, etc., and emphasizes adoption by enterprise-level teams.

Open source recognition: The number of stars and releases of the GitHub warehouse are both in a high range, indicating that it has stable attention and contributions in the developer community.

Undisclosed items: The number of independent paying customers, industry distribution and renewal rate have not been officially disclosed. Commercial details are subject to official subsequent disclosures.

Cost advantage

C-side/Individual: Open Source and Free Cloud are both $0. Free Cloud includes 25k spans/month and 60-day retention, which is suitable for low-threshold verification.

Developer/API Tier: Pro Cloud is $19/month, including 100k spans/month, and additional usage is marked at $5/100k spans.

Enterprise layer: Enterprise is Custom, supporting SSO, compliance and SLA. Additional retention periods, private deployment and support response terms should be verified at time of purchase.

Main functions

• End-to-end Trace and Debugging: Trace user interactions, context retrieval and tool call links.
• LLM-as-a-Judge Evaluation: Provides 30+ evaluation indicators, covering relevance, illusion, task completion, etc.
• Test Suites and Assertions: Supports framed execution of unit and regression tests.
• Production Monitoring and Alarm: Supports online evaluation rules, cost tracking and abnormal alarms.
• Optimization Toolchain: Includes Agent Playground and Prompt Optimizer.

Model and version evolution

Mainline version of Opik

2.0.58 (2026-06-08): The latest version, the release record contains changes related to Opik V2 default and performance optimization.

Recent versions of Opik

2.0.57 (2026-06-04): Continuous version nodes, focusing on SDK, front-end and back-end repairs.

2.0.56 (2026-06-03): Continuously release nodes, reflecting the weekly-level high-frequency iteration rhythm.

Technical advantages

Integration of observation and evaluation: The same platform covers Trace, evaluation and optimization, reducing the cost of switching between multiple tool chains.

Scale processing capability: The official design goal is 40M+ traces/day, which is targeted at production-level loads rather than just experimental environments.

Multiple deployment strategies: Supports both OSS self-hosting and cloud solutions, making it easy to switch according to compliance and team stages.

How to use

How to use	Entrance	Typical steps	Adaptation scenarios
Open Source	GitHub repository + opik.sh	Deploy local or K8s -> SDK access -> Create test set	High data boundary requirements and operation and maintenance capabilities
Free Cloud	Comet Signup	Registration -> Configuration Project -> Access SDK	Quick verification for individuals or small teams
Pro/Enterprise	Comet Pricing/Contact	Evaluate usage -> Configure team permissions -> Establish governance process	Scaled collaboration and compliance requirements

Product Pricing

Package	Public Price	Public Quota	Remarks
Open Source	$0	Self-hosted (no official public limit)	Core capabilities available
Free Cloud	$0	25k spans/month 60 days retention	Max 10 members
Pro Cloud	$19/month	100k spans/month with 60 day retention	Additional $5/100k spans
Enterprise	Custom	Custom	SSO, SLA, compliance enhancements

Application scenarios

• Agent debugging and regression: locate failed nodes in multi-step execution and reproduce the experiment.
• Evaluation-driven release: Use Test Suites and evaluation indicators for version access control.
• Production Quality Governance: Continuously track costs, error rates, and risk of policy violations.

Applicable people

• AI Application R&D Team: Tracking, evaluation, and optimization need to be integrated into the same process.
• Platform Governance Team: Need to support multi-project quality standards and audit traces.
• Enterprise Technology Leader: Focus on compliance SLAs and scalable deployment strategies.

Not suitable for the boundary: Teams that only require light log viewing, no continuous evaluation plan, or no governance process may not be able to realize the value of the platform in the short term.

Summary and Outlook

It provides competitive solutions in its field, and its core value lies in lowering the threshold for AI use in this field.

Current limitations: Some advanced features require paid subscription, and the free version has function or usage restrictions; specific technical details and performance benchmarks have not yet been fully disclosed.

Related tools: GitHub Copilot, Cursor

Version Info

• Opik 2.0.58 ：The latest version of GitHub Releases, release records show continued progress around Opik V2, evaluation and performance optimization. （2026-06-08）
• Opik 2.0.57 ：The mainline continuous version, the release record includes function fixes and integration updates of the front-end and SDK. （2026-06-04）
• Opik 2.0.56 ：Continuously releasing nodes shows that Opik maintains stability and availability under a high-frequency version rhythm. （2026-06-03）