Latitude
Open-source platform for building and operating LLM features.
About
Latitude is an open-source platform for building and running LLM features in production. You instrument your existing calls to get observability, then iterate in a prompt playground, curate datasets from real traffic, and run evaluations including LLM-as-judge, with a reliability loop that converts production failures into regression tests.
It is genuinely open source and self-hostable, and the staged adoption path (observe first, optimize later) is sensible. As a younger project, its ecosystem of templates and third-party integrations is thinner than the incumbents'.
What it can do
- LLM observability on real traffic
- Prompt playground with versioning
- Dataset curation from production
- Built-in and LLM-as-judge evaluations
- Prompt optimizer against eval suites
- OTLP-compatible telemetry ingest
Setup
- Follow the quick start at docs.latitude.so.
- Instrument your app with the Latitude Telemetry SDK.
- Open the prompt playground to reproduce and iterate.
- Curate datasets and wire evaluations into your release flow.
Prerequisites
An existing LLM integration to instrument, or a new project.
Use cases
Works with
Trust score — 6/10
Scored 6/10 (metrics tier, metrics-v1). Evidence: open-source prompt engineering platform; observability-first adoption path; younger project, thinner ecosystem. License: Open source, self-hostable.
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 5.0/10
- Maintenance · recency, releases, issue load
- 6.0/10
- Openness · license + docs
- 9.0/10
- Manual review · sanity read + red-flag check
- 6.0/10
- · open-source prompt engineering platform
- · observability-first adoption path
- · younger project, thinner ecosystem
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.