Spanlens is unclaimed —

Spanlens

One line. Every LLM call, observed.·spanlens.io

Spanlens is an open-source (MIT) LLM observability and monitoring platform that logs every OpenAI, Anthropic, and Gemini request via a one-line SDK swap or proxy baseURL change. It provides cost tracking, agent tracing with waterfall span trees, anomaly detection, PII and prompt-injection scanning, a model-savings recommender, evals, experiments, and user analytics. It is offered as a hosted SaaS or as a self-hostable Docker deployment, with the same codebase for both.

What it's for

Log every LLM request with model, tokens, cost, latency, and full bodyTrack per-request and daily cost with budget alertsTrace multi-step agent workflows as waterfall span treesDetect anomalies in latency or cost versus a rolling baselineScan requests for PII, prompt-injection patterns, and mask API keysRecommend cheaper models with dollar-figure savingsScore LLM responses with an LLM-as-judge (evals)Replay datasets across prompt versions and models (experiments)Analyze per end-user and per-session cost, volume, and errorsSelf-host the full platform on own infrastructure via Docker

Features 20

Agent tracing as waterfall span trees
Anomaly detection (3-sigma deviations vs 7-day baseline)
Automatic API key masking before storage
Budget alerts
Critical path highlighting in multi-step agent traces
Datasets / golden test sets from production traffic
Experiments to replay datasets across prompt versions and models
Human annotation queue for review
LLM-as-judge evals scoring responses 0 to 1
Model savings recommender with dollar-figure estimates
OpenTelemetry (OTLP/HTTP) ingest and export
Per-request cost breakdown and daily rollups
PII detection (SSN, credit card, email, IBAN, passport)
Prompt-injection pattern detection
Request logging with model, tokens, cost, latency, and full request/response body
Response caching via header or SDK helper
Self-hosting via one Docker Compose file
Versioned prompt library with diff, A/B rollout, and playground
Webhooks and Slack notifications
Weekly digest emails and data-silence alerts

At a glance

free tieropen sourceself-hostableAPI

TypeWeb app

DeploymentHybrid

PlatformsWeb, Linux, macOS, Windows

Forsolo developers, small teams, teams needing full visibility, large teams (Enterprise)

CompanyOceancode · 2026

Integrations

OpenAIAnthropicGeminiAzureBedrockVertexMistralLangChainLlamaIndexVercel AI SDKLangGraphSlackDiscordS3BigQueryOllamaOpenTelemetrySupabaseClickHousePaddle

Resources

Pricing

Enterprise pricing is custom/quote-based. Prices are stated in USD, exclusive of applicable taxes; VAT/GST is added at checkout by Paddle. Self-hosting via Docker is free with all features, no seat or usage cap.

FreeFree

For personal projects and exploration.

50K requests/month, then a hard 429; 14-day log retention

Pro$29/mo

For solo developers shipping to production.

100K requests/month, 3 seats, 2 workspaces, 90-day log retention

Team$149/mo

For teams that need full visibility.

1M requests/month, 10 seats, 5 workspaces, 365-day log retention

EnterpriseQuote

Custom pricing; contact sales.

Custom volume & rate limits, 365-day retention (extendable by contract), unlimited seats/workspaces