OpenLIT

license

Apache-2.0(OSI)

source

public

core-gated

ungated

Fully Apache-2.0, OpenTelemetry-native, no feature-gated core; vendor-neutral by design (exports to any OTel backend).

• https://github.com/openlit/openlit recorded 2026-08-13

  Repository page. License resolves to Apache-2.0 (spdxId "Apache-2.0"); the README strapline reads "OpenTelemetry-native LLM Observability, GPU Monitoring, Guardrails, Evaluations, Prompt Management, Vault, Playground" and the root listing includes docker-compose.yml with no ee/ or enterprise directory beside it. 2,686 stars.

openlit draws 414,172 PyPI downloads a month (97,995 in the last week, 20,112 in the last day), which lands in the 100K-1M band. For a hosted platform, the SDK's download count is the platform's reach, and the band follows the measured figure rather than a discretionary discount for the CI and transitive share of an OpenTelemetry auto-instrumentation library's installs. Roughly 2,700 GitHub stars still describe a smaller community than Langfuse or MLflow, and the level says so - 3 against Langfuse's 5.

• https://pypistats.org/api/packages/openlit/recent recorded 2026-08-14

  last_month = 414,172 downloads for openlit (last_week 97,995, last_day 20,112).

A broad feature matrix covering every dimension this category grades - OpenTelemetry-native tracing and metrics for LLMs, vector databases and GPUs, 11 built-in evaluation types for hallucination, bias and toxicity, a versioned Prompt Hub, cost tracking including custom and fine-tuned pricing, guardrails, a secrets vault and 50+ provider and framework integrations - plus GPU observability on top. That puts it level with Langfuse at 4. It is held there rather than 5 because no observability product is a clear frontier-definer over the MLflow and Langfuse anchors.

• https://github.com/openlit/openlit recorded 2026-08-13

  README strapline - "OpenTelemetry-native LLM Observability, GPU Monitoring, Guardrails, Evaluations, Prompt Management, Vault, Playground" - and a section headed "Observability, Evaluations, Rule Engine, Guardrails, Prompts, Vault, Playground", with integrations across 50+ LLM providers and vector databases.