PromptLayer

source

closed

license

Apache-2.0(OSI, client library MagnivOrg/prompt-layer-library only)

self-host

enterprise-only(commercially licensed Docker images and Helm charts, Enterprise plan only)

platform

proprietary-SaaS(API-key gated, hosted prompt registry/logging/evals)

The Apache-2.0 client library is open, but the prompt registry, logging and evaluation backend is closed. It runs as a hosted service, and a self-hosted deployment exists only on the Enterprise plan under a commercial license, so the product is closed rather than open core.

• https://docs.promptlayer.com/self-hosted recorded 2026-09-29

  Self-hosting page: the architecture is described as "fully dockerized services designed for scalability and reliability" with "pre-built Docker images with Helm charts" and a step to "discuss your requirements and obtain a license".

• https://github.com/MagnivOrg/prompt-layer-library recorded 2026-08-13

  Repository page. GitHub resolves its LICENSE to Apache-2.0 (spdxId "Apache-2.0"), and the repository description is "PromptLayer - Maintain a log of your prompts and OpenAI API requests. Track, debug, and replay old completions." It is a client library; the registry, logging and eval services it talks to are not in it and are reached over promptlayer.com with an API key.

• https://pypi.org/pypi/promptlayer/json recorded 2026-09-29

  PyPI metadata for the promptlayer package: summary "PromptLayer is a platform for prompt engineering and tracks your LLM requests." and classifier "License :: OSI Approved :: Apache Software License".

• https://www.promptlayer.com/pricing recorded 2026-09-29

  Pricing page: "Enterprise customers have the option of self-hosted on GCP, AWS, and Azure" while Free, Pro and Team plans are Cloud-hosted.

Adoption is measured by PyPI downloads of the promptlayer package. PromptLayer positions itself as a no-code prompt CMS for non-technical teams rather than a full observability platform, and no harder user or account count is disclosed beyond the download figure.

• https://pypistats.org/api/packages/promptlayer/recent recorded 2026-10-07

  91,871 downloads in the trailing 30 days for promptlayer (pypistats last_month)

PromptLayer covers the prompt lifecycle: a registry with versioning and a no-code editor, request logs and traces (including OpenTelemetry ingestion), and evaluations with LLM-as-judge and regression runs. It stays a step behind Langfuse on open, self-serve deployment and the breadth of framework instrumentation.

• https://docs.promptlayer.com recorded 2026-09-29

  Docs home: "requests, responses, metadata, cost, latency, and feedback in one timeline." and "Compare changes against real examples before they reach users."

• https://docs.promptlayer.com/features/evaluations/overview recorded 2026-09-29

  Evaluations page: "Add scoring columns such as LLM-as-judge, human grading, equality comparison, cosine similarity, or code evaluators."

• https://docs.promptlayer.com/features/observability/traces/opentelemetry recorded 2026-09-29

  OpenTelemetry page: "Any OpenTelemetry SDK or Collector can export traces to this endpoint." PromptLayer exposes an OTLP/HTTP endpoint for traces, and gen_ai model and provider attributes are parsed into request logs.

• https://github.com/MagnivOrg/prompt-layer-library recorded 2026-08-13

  Repository tagline - "PromptLayer - Maintain a log of your prompts and OpenAI API requests. Track, debug, and replay old completions." Prompt versioning and request logging are what the library offers; no tracing-platform, dataset or LLM-as-judge surface lives in it.