← all SaaS

Apache-2.0

34 self-hostable open-source tools licensed under Apache-2.0 · across 26 categories

Permissive like MIT, plus an explicit patent-grant clause and a notice file requirement. Common when the project has corporate maintainers who want explicit patent peace for downstream users.

One-line plain-English summary above — not legal advice. Confirm with the project's LICENSE file before commercial use.

AI code completion / chat in the IDE

  • Aider alive ★ 44.8k replaces GitHub Copilot
    5min `pip install aider-chat` · Free CLI; model cost is the line item — pair with Ollama / vLLM for $0 marginal cost.
  • Continue alive ★ 33.2k replaces GitHub Copilot
    10min — VS Code or JetBrains plugin + a local model · Free for the plugin; model server is the cost line — $0 on a workstation 16GB+ GPU, ~$200/mo on cloud GPU.
  • Tabby alive ★ 33.5k replaces GitHub Copilot
    20min docker run with --gpus + plugin install · $200+/mo cloud GPU for the inference server; $0 on a workstation. Plugin overhead is negligible.

API client / API workspace

  • Insomnia alive ★ 38.4k replaces Postman
    5min (desktop app) · $0 for local-only use; cloud sync requires Kong account.

App platform / git-push deploy

  • CapRover alive ★ 15.0k replaces Heroku
    20min (Docker Swarm setup) · $10 VPS for the controller.
  • Coolify alive ★ 55.0k replaces Heroku +1
    10min (single-line bash installer) · $10 VPS for the controller + apps.

Backend-as-a-service (auth + DB + storage + realtime)

  • Supabase alive ★ 102.3k replaces Firebase
    20min docker-compose (the official self-host bundle) · $10-20 VPS — Postgres + GoTrue + PostgREST + Realtime + Storage in one stack.

Business intelligence / dashboards / data exploration

  • Apache Superset alive ★ 72.8k replaces Tableau
    30min docker-compose (Superset + Postgres + Redis + Celery worker) · $15 VPS for the multi-container stack; production setups want a dedicated Postgres + Redis.

Cloud file sync + sharing for teams

Container image registry

  • Distribution (CNCF) alive ★ 10.4k replaces Docker Hub
    5min docker run · $3-10/mo VPS; storage on local disk or S3.
  • Harbor alive ★ 28.5k replaces Docker Hub
    30min — `installer.sh` with bundled compose (Harbor + Postgres + Redis + Trivy) · $10-30/mo VPS — storage scales with image volume; back with S3-compatible blob store.

Distributed tracing / observability

  • Jaeger alive ★ 22.8k replaces Honeycomb
    10min docker run (`all-in-one`) for dev; 1h for production with Cassandra/Elasticsearch backend · $5/mo VPS for dev; $50+/mo for production with Elasticsearch storage.

Document database (managed)

  • FerretDB stale ★ 10.9k replaces MongoDB Atlas
    10min docker-compose (FerretDB + Postgres) · $5-10 VPS — Postgres backend does the heavy lifting.

Feature flags / progressive delivery

  • Unleash alive ★ 13.5k replaces LaunchDarkly
    20min docker-compose (Node + Postgres) · $10/mo VPS for the app; SDKs are free.

Frontend hosting / serverless deploy

  • Dokploy alive ★ 34.0k replaces Vercel
    10min docker-compose · $10 VPS for the controller.

Hosted CI/CD

  • Concourse alive ★ 7.83k replaces CircleCI
    30min docker-compose (web + worker + Postgres) · $15-30/mo VPS — workers eat resources during builds.
  • Drone CI alive ★ 35.8k replaces CircleCI
    20min docker-compose (Drone server + at least one runner) · $10-20/mo VPS for the server; runners are separate (one $5 VPS per concurrent build).
  • Woodpecker CI alive ★ 7.02k replaces CircleCI
    20min docker-compose (server + runner) · $10/mo VPS for the server; runners on $5 boxes.

Identity / SSO / authentication-as-a-service

  • Keycloak alive ★ 34.4k replaces Auth0
    20-30min docker-compose (Keycloak + Postgres) · $10 VPS for a small team; $20+ for production with HA Postgres.
  • Zitadel alive ★ 13.8k replaces Auth0
    20min docker-compose · $10-15 VPS — Go binary + CockroachDB or Postgres.

Image / video CDN + on-the-fly transformation

Live streaming + chat

  • Restreamer stale ★ 5.03k replaces Twitch
    15min docker run · $5-10/mo VPS; bandwidth-bound for viewer count.

LLM inference API

  • vLLM alive ★ 79.9k replaces OpenAI API
    30min docker run with --gpus · $200-1500/mo depending on GPU class; an A100 80GB runs Llama 3.1 70B comfortably with PagedAttention batching.

Message queue / async messaging

  • NATS JetStream alive ★ 19.8k replaces Amazon SQS
    5min single-binary or docker run · $5-10/mo VPS — Go binary is light; clustering is straightforward.

Object storage / blob store

  • SeaweedFS alive ★ 32.3k replaces AWS S3
    15min docker-compose (master + volume + filer) · $5-10 VPS for small deployments; designed to scale to billions of small files cheaply.

Online whiteboard / visual collaboration

  • drawio (diagrams.net) alive ★ 5.37k replaces Miro
    5min docker run · $3/mo VPS; static frontend.
  • tldraw alive ★ 47.0k replaces Miro
    10min — clone the example sync server + frontend · $5-10/mo VPS for the sync server.

Public status page / incident communication

RSS feed reader

  • Miniflux alive ★ 9.22k replaces Feedly
    10min docker-compose (Go binary + Postgres) · $3 VPS easily handles a personal install with hundreds of feeds; Postgres can be tiny.

Search-as-a-service

  • OpenSearch alive ★ 12.9k replaces Algolia +1
    30min docker-compose · $20+ VPS — JVM-heavy; want 4GB RAM dedicated.

Team chat

  • Zulip alive ★ 25.2k replaces Slack
    30-60min (their installer script on Ubuntu 22.04) · $10 VPS minimum (their installer wants 2GB+ RAM).

Video conferencing

  • Jitsi Meet alive ★ 29.2k replaces Zoom
    15-30min (Debian package + nginx) · $10-20 VPS depending on concurrent meeting size; CPU-bound during calls.

Website builder + hosting

  • Hugo + Decap CMS alive ★ 88.0k replaces Squarespace
    30min — Hugo install + theme + Decap CMS for editor UI · $0-5 — Hugo builds static HTML; deploy free on Cloudflare Pages or Netlify.

Writing assistant / grammar + style checker

  • Harper alive ★ 10.5k replaces Grammarly
    5min — Rust binary, also runs as a VSCode extension and a browser extension · $0 — Harper runs locally (LSP) or in-browser (WASM) by default; no server required for personal use.