Model Config & .env EOL Auditor
0ms Browser-Local · Zero Network Upload
Paste .env variables, litellm_config.yaml router blocks, CSV rows, or plain model IDs to flag retired and expiring models locally in your browser—complete with provider-isolated matching and 1-click DevToken replacement cost checks.
2. Lifecycle & Replacement Diagnostics
| Config Key | Ecosystem | Model ID | 4-State Status | Match Type | Shutdown Date | Replacement & Docs |
|---|
Provider-Isolated Matching & Privacy Guarantee
Every line you paste is parsed in memory inside your browser tab; API keys, comments, and non-model environment variables are ignored and never transmitted over the network. Our matcher strips AWS Bedrock regional inference prefixes (us., eu., apac., global.) and -vN:M suffixes for alias lookup while preventing snapshot date collisions across dated versions. Native OpenAI model IDs absent from OpenAI's official deprecation table report ACTIVE (no_eol_plan) rather than borrowing Azure OpenAI retirement dates.
What configuration formats does the local auditor recognize?
You can paste shell .env assignments, litellm_config.yaml model entries (openai/..., bedrock/..., vertex_ai/..., gemini/..., azure/...), JSON key-value pairs, two-column provider,model_id CSV rows, or newline-separated model IDs.
What is the difference between no_eol_plan and unlisted match types?
Vendors like OpenAI and Google Gemini publish deprecation tables rather than closed finite model lists, so valid model IDs absent from their deprecation page resolve to no_eol_plan. Catalog-backed ecosystems (Anthropic, Vertex AI, AWS Bedrock, Azure OpenAI) include full supported model inventories, so unrecognized strings surface unlisted to help catch typos or custom fine-tune names.