Skip to the content
Know your tech before you buy it. Makers: get listed from $49/mo

Odigos

odigos.io

  • Free plan
  • Windows
  • macOS
  • Linux
  • Self-hosted
7.5 Club score Visit odigos.io

What Odigos is

Odigos records live runtime activity inside services so engineers, AI agents, and security teams can investigate production behavior without changing application code. It collects traces, metrics, logs, profiles, and function values, including evidence that might otherwise call for an instrumentation release or redeployment. Its out-of-process eBPF capture is stated to use under 1% CPU and add effectively zero latency. Teams can control capture access and scope with RBAC and policies, then shape values in the cluster using PII masking, attribute deletion, and sampling. Odigos exports data as OpenTelemetry and can run alongside tools including Datadog, New Relic, Honeycomb, Grafana Cloud, Jaeger, Tempo, Loki, and SigNoz. Deployments are described for Kubernetes, VMs, and bare metal; listed languages include Go, Java, Python, .NET, JavaScript, PHP, and Ruby. The self-hosted Open Source plan is listed as free with community support. Enterprise pricing is custom and includes a 14-day free trial; Oracle database tracing is marked “Coming Soon.”

Who it suits

Odigos is for engineers, AI agents, and security teams investigating production behavior, containing attacks, or improving code. It suits teams seeking self-hosted runtime visibility and controls over what gets captured.

Good to know, and what to watch

  • Captures multiple runtime signals without an instrumentation release.
  • Supports function-level virtual patching before a code release.
  • Offers in-cluster masking, deletion, and sampling controls.
  • Exports OpenTelemetry to configured destinations.
  • Open Source plan includes community support.
  • Oracle database tracing is marked “Coming Soon.”
  • Enterprise pricing is custom.
  • Enterprise support is limited to listed 24/7 premium support.

Enterprise

—

Open Source

Free

Free plan · 14-day trial

Getting started

  1. Open the Odigos website
  2. Choose the Open Source plan or start the Enterprise 14-day free trial
  3. Deploy Odigos on Kubernetes, a Linux VM or cloud VM, or Linux bare metal
  4. Configure capture policies and access controls
  5. Select OpenTelemetry destinations for exported data

Limits to check

The Open Source plan is free and provides core OpenTelemetry tracing with community support. Enterprise adds database tracing, multi-cluster administration, and 24/7 premium support, but Oracle database tracing is marked Coming Soon.

The Club's review

Odigos is a self-hosted runtime observability tool for teams investigating live services, especially when existing telemetry leaves gaps. Its broad capture and security controls are useful without a code release, but Enterprise administration and support carry custom pricing.

Overview

Odigos collects traces, metrics, logs, profiles, and function values from running services without requiring application-code changes or redeployment. It exports data as OpenTelemetry to configured destinations and can run alongside established tools such as Datadog, New Relic, Honeycomb, Grafana Cloud, Jaeger, Tempo, Loki, and SigNoz. That makes it an addition to an existing stack rather than a reason to replace one. Its self-hosted model suits teams that want runtime capture in their own environment, but also means deployment and operation remain with the team.

Key features

  • Live eBPF capture: Odigos says its out-of-process capture uses under 1% CPU and adds effectively zero latency. The approach is appealing when production evidence is missing and an instrumentation release is not practical; the performance figures are the maker’s claims.
  • Function-level virtual patching: Teams can block matching application calls before a code release. That gives security teams a way to contain a problem while a permanent code fix is prepared, but it is not a replacement for fixing the underlying issue.
  • Capture governance and data handling: RBAC and policy controls govern who can request captures, which workloads are in scope, and what may be captured. PII masking, attribute deletion, and sampling can shape values in the cluster before export. These controls are important because runtime capture can expose sensitive data; teams still need to decide and enforce suitable policies.
  • Language and environment coverage: The pricing page names Go, Java, Python, .NET, JavaScript, PHP, and Ruby. Odigos describes installation on Kubernetes, VMs, and bare metal; the plan comparison specifies tracing for Linux VMs or cloud VMs and Linux bare metal. Oracle database tracing is marked “Coming Soon,” a material gap for teams whose investigations depend on that database.
  • Operational safeguards: Odigos says it is SOC 2 audited. Open Source includes community support, while Enterprise lists 24/7 premium support—an important distinction for teams that need a response commitment.

Pricing

Odigos uses a freemium model with a free plan and a 14-day Enterprise trial. The self-hosted deployment model applies across the offering.

  • Open Source — free: Free and open source under Apache 2.0, with core OpenTelemetry tracing and community support. It is the sensible starting point for teams that need core tracing and can operate without Enterprise administration or premium support. It gives up the Enterprise plan’s database tracing, multi-cluster administration, and 24/7 support.
  • Enterprise — custom pricing: Includes low-overhead eBPF capture, database tracing, multi-cluster administration, and 24/7 premium support, with a 14-day free trial. It fits teams that need those operational capabilities or premium support; Oracle tracing remains marked “Coming Soon.”

No seat or usage quota is stated for either plan, so the practical choice is driven by required capabilities and support rather than a published capacity threshold.

Platforms

Odigos lists Linux, macOS, Windows, and self-hosted platforms, with Kubernetes support, network visibility, application tracing, and kernel profiling. Its deployment descriptions focus on Kubernetes, Linux VMs or cloud VMs, and Linux bare metal, so teams should match their target environment to those stated deployment options.

Who it's for

Odigos is aimed at engineers, AI agents, and security teams investigating production behavior, containing attacks, or improving code. It is strongest for organizations that need evidence from live workloads without waiting for an instrumentation release, and that can manage a self-hosted system. It is less suitable when the requirement is Oracle database tracing today, or when a team needs Enterprise-grade administration and support without moving to custom pricing.

Pros and cons

  • Pro: Captures five kinds of runtime signal without code changes or redeployment, helping teams investigate gaps in existing telemetry.
  • Pro: OpenTelemetry export and compatibility alongside named observability platforms let teams add capture without replacing their current stack.
  • Pro: RBAC, capture policies, and in-cluster data shaping give teams controls over sensitive runtime evidence.
  • Con: Multi-cluster administration and 24/7 premium support are Enterprise features with custom pricing, not part of the free Open Source plan.
  • Con: Oracle database tracing is not ready, which limits the fit for teams that need it for production investigations.
  • Con: Self-hosting puts deployment and operation on the customer rather than offering a managed service.

Alternatives

Consider eBPF Observability Tools to compare tools in the category. DeepFlow is a free, open-source option with a Community Edition at no cost and support for Linux servers and selected Kubernetes, cloud, and container environments. Coroot offers a self-hosted Apache 2.0 Community Edition at no cost and a Standard plan at $1 a month, making it an alternative to compare if those plan terms suit your needs. groundcover has a free-forever BYOC plan at no cost with 12-hour data retention, so it is worth considering when that retention window meets the use case. Metoro Kubernetes Profiling offers a Hobby plan at no cost for one cluster, one user, two nodes, and 28-day retention; it fits a small Kubernetes setup within those caps. Kubeshark has a Community plan at no cost, capped at three nodes or 60 pods and requiring internet connectivity, so it is an option when those limits are acceptable. Elastic Observability is a paid, usage-based alternative whose logs and traces ingest starts at $0.09/. Qpoint offers a Community plan at no cost for up to 25 endpoints, with agent discovery and monitoring with attribution. Caretta is a free, self-hosted option for Linux.

Verdict

Choose Odigos if your team needs live runtime evidence across multiple signal types, wants to retain its current observability stack, and can operate a self-hosted deployment. Its strongest case is adding capture without code changes, with controls for sensitive data and a virtual-patching response path. Look elsewhere if Oracle database tracing is essential now, or if you need multi-cluster administration and premium support without a custom-priced Enterprise plan.

The verdict

Odigos focuses on runtime evidence and capture controls without requiring code changes. Its self-hosted model and custom-priced Enterprise plan are worth weighing alongside the database tracing limitation.

Questions people ask

Is Odigos free?

Yes. The Open Source plan costs nothing and is free and open source under Apache 2.0.

Does Odigos offer a trial?

Enterprise includes a 14-day free trial.

Where can Odigos run?

It is self-hosted and can be installed on Kubernetes, Linux VMs or cloud VMs, and Linux bare metal.

Which languages are supported?

The pricing page lists Go, Java, Python, .NET, JavaScript, PHP, and Ruby.

What support is included?

Open Source provides Odigos community support. Enterprise lists 24/7 premium support.

Can it work with other observability tools?

Odigos exports OpenTelemetry and can run alongside Datadog, New Relic, Honeycomb, Grafana Cloud, Jaeger, Tempo, Loki, and SigNoz.

What the maker publishes

Access controls
Odigos says RBAC and policy controls govern who can request captures, which workloads they apply to, and what may be captured. source
Compliance
Odigos says it is SOC 2 audited. source
Data handling
PII masking, attribute deletion, and sampling can shape captured values in the cluster before export. source
Deployment
The company describes installation on Kubernetes, VMs, and bare metal; its plan comparison specifies Linux VM/cloud VM and Linux bare metal tracing. source
Destinations
Odigos exports data as OpenTelemetry to configured destinations and can run alongside Datadog, New Relic, Honeycomb, Grafana Cloud, Jaeger, Tempo, Loki, and SigNoz. source
Enterprise limits
Oracle database tracing is marked “Coming Soon” on the plan comparison. source
Instrumentation
Odigos captures missing runtime evidence live without requiring an instrumentation release or redeploy. source
Intended users
Odigos describes its users as engineers, AI agents, and security teams investigating production, containing attacks, and improving code. source
Languages
The pricing page lists Go, Java, Python, .NET, JavaScript, PHP, and Ruby as supported languages. source
Performance
Odigos states that its out-of-process eBPF capture uses under 1% CPU and adds effectively zero latency. source
Purpose
Odigos records live runtime activity inside services so engineers, AI agents, and security teams can investigate production behavior without changing application code. source
Security response
Odigos supports function-level virtual patching to block matching application calls before a code release. source
Signals
Odigos can collect traces, metrics, logs, profiles, and function values. source
Support
The open source plan provides Odigos community support, while Enterprise lists 24/7 premium support. source

Checked against the maker's pages on 29 Sept 2026.