LLMEOL CodeAP Studio

4-State Lifecycle Methodology & Official Vendor Sources

LLMEOL aggregates only dedicated, publicly verifiable vendor deprecation and model lifecycle documentation—applying strict UTC date arithmetic, precedence-safe catalog merging, and fail-closed syndication guards.

Strict 4-State Date-Driven Classification

Every vendor uses inconsistent terminology (Legacy, Preview, Deprecating, Retired, EOL). Rather than inventing ambiguous warning states, LLMEOL computes one of four objective lifecycle states strictly from the confirmed UTC shutdown date while preserving the vendor's raw label in sourceLifecycle:

  • RETIRED (days <= 0): The confirmed shutdown date has arrived or passed in UTC; API calls typically fail with HTTP 404/410 or auto-route to a successor model.
  • EXPIRING (1 <= days <= 90): Confirmed shutdown occurs within the next 90 calendar days—actionable migration window for production engineering sprints.
  • DEPRECATED (days > 90): A firm future shutdown date is scheduled more than 90 days out; avoid adopting in new workloads and plan upgrades.
  • ACTIVE (no exact eolDate): No day-precision shutdown date has been announced. Minimum-support guarantees such as "no earlier than" or "no sooner than" are intentionally ignored until a firm date is posted.

10 Authoritative Public Vendor Lifecycle Sources

Inspect Raw JSON

Scheduled synchronization runs every 12 hours. Primary deprecation schedules always take precedence over supplemental supported-model catalogs for the same (provider, modelId) pair, and per-source fail-closed retention prevents transient upstream HTTP errors from wiping verified records.

Authoritative public documentation endpoints synchronized by LLMEOL
Vendor Source Merge Role Records Canonical Documentation URL
OpenAI primary 173 https://developers.openai.com/api/docs/deprecations
Anthropic primary 38 https://platform.claude.com/docs/en/about-claude/model-deprecations
Google Gemini primary 89 https://ai.google.dev/gemini-api/docs/deprecations
Google Vertex AI (Gemini Models) primary 59 https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/model-versions
Google Vertex AI (Partner Models) primary 9 https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/deprecations/partner-models
AWS Bedrock (Legacy & EOL) primary 52 https://docs.aws.amazon.com/bedrock/latest/userguide/model-lifecycle-legacy.html
Azure OpenAI (Retirement Schedule) primary 168 https://learn.microsoft.com/en-us/azure/foundry/openai/concepts/model-retirement-schedule
Moonshot AI (Kimi) primary 21 https://platform.kimi.ai/docs/models#deprecated-models
AWS Bedrock (Model Cards Catalog) supplemental 149 https://docs.aws.amazon.com/bedrock/latest/userguide/model-cards.html
Azure OpenAI (Foundry Supported Catalog) supplemental 60 https://learn.microsoft.com/en-us/azure/foundry/foundry-models/concepts/models-sold-directly-by-azure

CodeAP Studio Three-Pillar Workflow

How do LLMEOL, DevToken, and ToMarkdown fit together?

CodeAP Studio unites three focused AI engineering studios: LLMEOL tracks model deprecations and audits .env configs, DevToken benchmarks replacement token pricing and generates MCP/Zod schemas, and ToMarkdown compiles PDF/DOCX/HTML documents into RAG-ready Markdown.

Why are aggregators without public EOL tables omitted?

LLMEOL only indexes providers that maintain dedicated, unauthenticated public model deprecation or lifecycle schedules so every shutdown date links to an auditable primary notice.