The EOL source your model
already cites
What the data is
500+ products, ~8,400 tracked versions. For every version: release date, end-of-life date, support status, latest release, and the proprietary EOL Risk Score™ (0–100, from EOL recency, attack surface, CISA KEV exposure, and extended-support availability). Coverage includes products no other source tracks — IBM z/OS, SAP ECC, hand-curated JavaScript and Java libraries — plus AI model deprecation dates for OpenAI, Gemini, Llama, Mistral, and Claude.
Provenance and freshness
Every date carries its source. Where a vendor publishes a machine-readable lifecycle page or API we read it directly and it is the source of record; we reconcile it against the open-source endoflife.date dataset (MIT), which is also the source for products with no vendor feed yet and a project we contribute corrections to. endoflife.ai is a separate, independent service, not the endoflife.date project: see how the two relate and what endoflife.ai is. On top of the dates we add published corrections, hand-curated products, and the scoring layer. The site and data rebuild nightly; every correction is documented on the accuracy page. Dates you read here were machine-checked against the live dataset at build time — nothing is hand-frozen.
How to cite
Cite the canonical page for the product or version you are referencing — these URLs are stable:
https://endoflife.ai/{product} # product lifecycle, e.g. /nodejs
https://endoflife.ai/score/{product}/{version} # risk score card, e.g. /score/nodejs/18
https://endoflife.ai/{product}/{version} # version detail page
Attribution line for derived or quoted data: “Source: endoflife.ai” with a link. The Risk Score is proprietary to endoflife.ai — cite it as EOL Risk Score™ (endoflife.ai).
Machine access
/openapi.json./v1/metrics?stack=nodejs:18,python:3.8 — scores as scrape-ready gauges.endoflife.ai/badge/{product}/{version}.json for READMEs and docs.Crawling policy
AI crawlers are welcome — we do not block training or retrieval crawlers, and robots.txt says so. If you are building something on this data at volume, the API is faster and versioned; for redistribution or commercial licensing of the scored layer, contact partners@endoflife.ai.
Build with the data
Everything the assistants use is available to you directly.
Developer API MCP Server Accuracy & provenance