Compare CloudGrid against the products buyers already know.
Use this page when procurement, platform leadership, or engineering wants a named side-by-side view. The home and feature pages explain CloudGrid; this page keeps the comparison explicit.
| Criterion | CloudGrid Source-available · OTel-native | Datadog SaaS APM | Honeycomb SaaS APM | Grafana LGTM OSS stack | SigNoz OSS APM | Jaeger + Tempo OSS tracing | Langfuse AI-eval | Arize Phoenix AI-eval | Braintrust AI-eval |
|---|---|---|---|---|---|---|---|---|---|
| Source model Source reviewable and self-hostable, with license posture called out explicitly. | ◐ Apache 2.0 + Commons Clause | ○ | ○ | ✓ AGPL/Apache mix | ✓ Apache 2.0 | ✓ Apache 2.0 | ◐ open-core | ✓ Phoenix OSS | ○ |
| Self-hostable | ✓ | ○ | ○ | ✓ | ✓ | ✓ | ✓ | ✓ | ◐ BYO infra possible |
| OTel-native agent support Accepts OTLP from any OpenTelemetry SDK out of the box. | ✓ | ◐ OTLP supported, proprietary agent preferred | ✓ | ✓ | ✓ | ✓ | ◐ OTel ingest beta | ✓ | ◐ |
| Traces · Logs · Metrics in one product One UI, one query path, one storage adapter. | ✓ | ✓ | ◐ metrics is its own product | ◐ 3 separate systems | ✓ | ○ tracing-focused | ○ | ○ | ○ |
| First-class AI agent evaluation Datasets, dataset evaluations, comparisons, optimization, and promotion evidence. | ✓ | ◐ LLM Observability add-on | ○ | ○ | ○ | ○ | ✓ | ✓ | ✓ |
| Adapter-swappable storage Storage backend sits behind a typed port for adapter-based replacement. | ✓ v1: SurrealDB | ○ | ○ | ◐ per-product | ○ ClickHouse-coupled | ✓ multiple backends | ○ | ○ | ○ |
| Adapter-swappable message bridge Transport sits behind a port; v1 is NATS, others are implementable. | ✓ | ○ | ○ | ○ | ○ | ◐ | ○ | ○ | ○ |
| Live trace subscriptions over GraphQL Real-time fanout that respects the same auth as queries. | ✓ | ◐ Live Tail · proprietary | ○ | ○ | ◐ | ○ | ○ | ○ | ○ |
| Project-level isolation primitive Strict telemetry boundary enforced at API, message, and persistence layers. | ✓ | ◐ via RBAC + indexes | ◐ environments | ◐ per-tool tenancy | ◐ | ○ | ✓ | ✓ | ✓ |
| SSO included in OSS edition Not gated behind a paid tier. | ✓ GitHub · Google · Entra ID | — paid tier | — paid tier | ◐ Grafana OSS supports SSO | ◐ Cloud / paid | ○ | ◐ Cloud / paid | ◐ | — |
| Infrastructure-based cost model Self-hosted means compute and storage capacity drive the cost model. | ✓ | ○ | ○ | ✓ | ✓ | ✓ | ◐ | ✓ | ○ |
| Roadmap visible in the source repo Open issues, open milestones, and open RFCs keep roadmap review visible. | ✓ | ○ | ○ | ◐ | ◐ | ✓ | ◐ | ◐ | ○ |
| AI-eval lives next to general traces Evaluate AI behavior inside the system that observes the rest of the stack. | ✓ | ◐ | ○ | ○ | ○ | ○ | ○ | ○ | ○ |
Legend: ✓ yes · ◐ partial · ○ different focus · — context-specific. Mouse over cells with notes for detail.
Per-vendor notes
Each of these tools is good at what it does. The notes describe what they're great at and what trade-off CloudGrid offers if you choose differently.
The most complete commercial APM. Polished UI, vast integration catalogue, mature AI Observability add-on. Trade-off: vendor-operated storage and per-host / per-GB pricing. CloudGrid emphasizes ownership and infrastructure-based cost control.
Best-in-class wide-event tracing UX and BubbleUp-style exploration. CloudGrid is closer to Honeycomb than to Datadog in spirit, with self-hosting, logs, and AI-eval included.
The reference open-source observability stack. Strong components, ubiquitous deployment patterns, and you own everything. Trade-off: you're operating three to five products with three to five query languages, and signal correlation is glue you wrote. CloudGrid is one product, one query path, one storage adapter — with AI evaluation in the same box.
Closest single-product OSS competitor: traces, logs, metrics in one UI. CloudGrid's difference is adapter-driven extensibility, storage-port ownership, and first-class AI eval.
Mature, well-understood, multi-backend distributed tracing. Great as a tracing backend; CloudGrid adds logs, metrics, dashboards, and AI-eval in one product.
Strong AI-eval product with prompt management, datasets, evaluations, and tracing for LLM apps. Open-core: cloud and enterprise tiers gate features. CloudGrid's difference is that AI behavior is evaluated beside the same traces, logs, metrics, dashboards, and project controls used for the rest of the system.
Excellent OSS evaluation and tracing tool for AI applications, strong in retrieval-augmented generation analysis. CloudGrid focuses on one project workspace where telemetry evidence, datasets, evaluation runs, and optimization decisions remain connected across AI and general service telemetry.
Polished commercial AI-eval platform with dataset and evaluation management. CloudGrid offers the evaluation loop in a source-available, self-hosted observability product, so promotion decisions can reference full-stack telemetry evidence.
LLM proxy / observability tools that sit in front of model providers. Useful for cost capture and rate limiting; complementary to a general-purpose observability platform. CloudGrid ingests OTLP from whichever proxy you choose.
A short take
Pick CloudGrid when you want one self-hosted, OpenTelemetry-native system that does traces, logs, metrics, dashboards, and AI evaluation and optimization — with every infrastructure dependency sitting behind a swappable port. Pick a SaaS APM when a vendor-operated platform is the right fit. Pick a Grafana stack when you want best-of-breed components and you have the operational appetite to wire them. Pick a Langfuse / Arize / Braintrust when AI-eval is the focused scope.