Model Deprecation Tracker

Track deprecated LLM endpoints and upcoming end-of-life dates. Know what to migrate before your API calls start failing.

Deprecation Tracker
ModelProviderDeprecatedReplacementAction
GPT-4 (0314)OpenAI2024-06-13gpt-4oMigrate now
GPT-3.5 Turbo (0301)OpenAI2024-09-13gpt-5-miniMigrate now
Claude 2Anthropic2025-03-01claude-haiku-4-5Migrate now
Claude Instant 1Anthropic2025-03-01claude-haiku-4-5Migrate now
Gemini 1.5 ProGoogle2025-09-24gemini-2-5-proMigrate now
GPT-4 32kOpenAI2025-06-06gpt-5-4Migrate now

Deprecation dates from official provider announcements. Estimated EOL dates are editorial estimates, not official announcements. Dataset last verified 2026-08-20. Active models tracked: 105.

How to Use

Check deprecated models

See models that have already passed their end-of-life date. If you're calling these, your API calls will fail or redirect unexpectedly.

Review upcoming EOLs

See models with estimated or announced end-of-life dates. Plan your migration at least 4–6 weeks ahead.

Find the replacement

Each deprecated model shows the recommended replacement. Check pricing and context window differences before migrating.

Update your code

Replace model ID strings in your codebase. Use model aliases ("model-name-latest") where available to avoid future manual updates.

Migration checklist

Before migrating from a deprecated model: (1) Identify all places in your codebase where the model ID is used. (2) Review pricing differences — the replacement may cost more or less. (3) Check context window and output limit changes. (4) Run your eval suite on the replacement. (5) A/B test in production before full cutover. (6) Update model aliases in config, not hardcoded strings.

FAQ

Have more questions? Contact us

Last verified: 2026-08-20 · methodology · data sources