← All comparisons

Spanlens vs Comet Opik · 2026

Evaluation-first against cost-first. Both fully open.

Summary

Opik and Spanlens are both properly open source with nothing held back, so the licence argument that separates other tools does not apply here. They differ in what they are organised around. Opik starts from output quality: datasets, judges, and experiments are the main surface, and instrumentation is SDK-first. Spanlens starts from the bill and the trace: one base URL captures existing calls, cost is attributed per request, and evaluation sits beside it. Opik is much larger and its evaluation workflow is deeper.

At a glance: Spanlens vs Comet Opik (2026)

Side-by-side feature comparison of Spanlens and Comet Opik in 2026.
FeatureSpanlensComet Opik
LicenceMITApache 2.0
Gated enterprise directory in the repoNoNo
Self-host the whole productYesYes
Single-command Docker installYesPartial
GitHub stars (2026-07-30)1020,966
One-line baseURL swapYesNo
Works without touching call sitesYesNo
SDK decoratorsPartialYes
OpenTelemetry (OTLP) ingestYesYes
Framework-agnosticYesYes
Per-request USD in the logYesYes
Dated model variant pricingYesPartial
Cost by prompt versionYesPartial
Cost by end user and sessionYesPartial
Cheaper-model recommendations with a dollar figureYesNo
Prompt-caching savings reportYesNo
LLM-as-judge scoringYesYes
Datasets of test casesYesYes
Offline experiments across versionsYesYes
Human annotationYesYes
Judge-to-human correlation as a metricYesPartial
Live traffic A/B with significance testYesNo
Depth of the evaluation workflow overallPartialYes
Agent trace waterfallYesYes
Critical path markedYesNo
Statistical anomaly detectionYesNo
Log durability queue on analytics-store failureYesNo
PII and prompt-injection scanningYesPartial
Response caching at the proxyYesNo

Updated 2026-07-30. Scroll for the grouped view with notes below.

Why teams pick Spanlens over Comet Opik

No call sites to change

Opik instruments your code: you add decorators or wrap clients, and every place that calls a model needs touching. Spanlens sits in front of the provider, so you change one base URL and existing calls are captured wherever they live. In a codebase with model calls spread across services, that difference decides how long adoption takes.

Cost is the organising idea

Spanlens computes per-request USD as the response passes through, matched to the exact dated model id, and rolls it up by model, prompt version, end user, and session. Opik tracks token usage and cost, but the product is built around scoring quality rather than around explaining a bill.

A savings recommender that names a number

Spanlens looks at your real token distribution per route and flags where a smaller model would plausibly hold quality, with an estimated monthly saving and a confidence tier. That is a different question from "did this response score well".

Critical path on agent traces

Both tools render multi-step runs as trees. Spanlens marks the longest dependency chain, so a five-tool agent tells you which span to fix instead of leaving you to compare timings by eye.

Anomaly detection and log durability

Spanlens flags 3-sigma deviations in latency, cost, and error rate against a rolling 7-day baseline per provider and model, and queues log writes to Postgres if ClickHouse is unavailable so rows are not silently dropped. These are operations concerns rather than evaluation concerns.

Feature-by-feature

Licence and hosting
Feature
Spanlens
Comet Opik
Licence
MIT
Apache 2.0
Gated enterprise directory in the repo
Self-host the whole product
Single-command Docker install
Opik documents Docker Compose and Kubernetes paths.
GitHub stars (2026-07-30)
10
20,966
Getting data in
Feature
Spanlens
Comet Opik
One-line baseURL swap
Works without touching call sites
SDK decorators
OpenTelemetry (OTLP) ingest
Framework-agnostic
Cost
Feature
Spanlens
Comet Opik
Per-request USD in the log
Dated model variant pricing
Cost by prompt version
Cost by end user and session
Cheaper-model recommendations with a dollar figure
Prompt-caching savings report
Evaluation
Feature
Spanlens
Comet Opik
LLM-as-judge scoring
Datasets of test cases
Offline experiments across versions
Human annotation
Judge-to-human correlation as a metric
Live traffic A/B with significance test
Spanlens reports a Welch t-test on latency and cost plus a z-test on error rate.
Depth of the evaluation workflow overall
Traces and operations
Feature
Spanlens
Comet Opik
Agent trace waterfall
Critical path marked
Statistical anomaly detection
Log durability queue on analytics-store failure
PII and prompt-injection scanning
Response caching at the proxy

Last updated 2026-07-30 · Spot something inaccurate? Let us know.

When Comet Opik might be the better fit

We don't think every team should pick us. Here's where Comet Opik legitimately wins.

Evaluation is deeper, and that is the point

If your central question is whether output quality is improving, Opik is built for that question first. Datasets, judges, experiment comparison, and the workflow around them are more developed than what Spanlens ships, and Spanlens evaluation exists alongside cost rather than as the centre of the product.

A much larger and faster-moving project

Opik has 20,966 GitHub stars against Spanlens at 10, and recorded more than 300 commits between 1 May and 30 July 2026, verified on 30 July. It is one of the fastest-growing tools in this category.

Apache 2.0 with nothing held back

Opik has no gated enterprise directory, so self-hosting gives you the whole repository. Spanlens is MIT with the same property, so on licence terms this is a tie rather than an advantage for either side.

Backed by an established company

Opik comes from Comet, which has been selling experiment tracking to ML teams for years. If vendor longevity is part of your decision, that history counts for something Spanlens cannot match yet.

Frequently asked questions

Why pick Spanlens over Comet Opik for "No call sites to change"?

Opik instruments your code: you add decorators or wrap clients, and every place that calls a model needs touching. Spanlens sits in front of the provider, so you change one base URL and existing calls are captured wherever they live. In a codebase with model calls spread across services, that difference decides how long adoption takes.

Why pick Spanlens over Comet Opik for "Cost is the organising idea"?

Spanlens computes per-request USD as the response passes through, matched to the exact dated model id, and rolls it up by model, prompt version, end user, and session. Opik tracks token usage and cost, but the product is built around scoring quality rather than around explaining a bill.

Why pick Spanlens over Comet Opik for "A savings recommender that names a number"?

Spanlens looks at your real token distribution per route and flags where a smaller model would plausibly hold quality, with an estimated monthly saving and a confidence tier. That is a different question from "did this response score well".

Why pick Spanlens over Comet Opik for "Critical path on agent traces"?

Both tools render multi-step runs as trees. Spanlens marks the longest dependency chain, so a five-tool agent tells you which span to fix instead of leaving you to compare timings by eye.

Why pick Spanlens over Comet Opik for "Anomaly detection and log durability"?

Spanlens flags 3-sigma deviations in latency, cost, and error rate against a rolling 7-day baseline per provider and model, and queues log writes to Postgres if ClickHouse is unavailable so rows are not silently dropped. These are operations concerns rather than evaluation concerns.

When is Comet Opik a better fit than Spanlens for "Evaluation is deeper, and that is the point"?

If your central question is whether output quality is improving, Opik is built for that question first. Datasets, judges, experiment comparison, and the workflow around them are more developed than what Spanlens ships, and Spanlens evaluation exists alongside cost rather than as the centre of the product.

When is Comet Opik a better fit than Spanlens for "A much larger and faster-moving project"?

Opik has 20,966 GitHub stars against Spanlens at 10, and recorded more than 300 commits between 1 May and 30 July 2026, verified on 30 July. It is one of the fastest-growing tools in this category.

When is Comet Opik a better fit than Spanlens for "Apache 2.0 with nothing held back"?

Opik has no gated enterprise directory, so self-hosting gives you the whole repository. Spanlens is MIT with the same property, so on licence terms this is a tie rather than an advantage for either side.

When is Comet Opik a better fit than Spanlens for "Backed by an established company"?

Opik comes from Comet, which has been selling experiment tracking to ML teams for years. If vendor longevity is part of your decision, that history counts for something Spanlens cannot match yet.

If your first question is whether quality is improving, Opik is built for it and is the safer pick on maturity. If your first question is what a request cost and where the latency went, and you would rather not instrument every call site, try Spanlens.

Free tier · No credit card · Self-host with Docker