Model retirements
When a model maker retires a model, requests to it start failing (or are quietly sent to another model). Here is every announced shutdown for the models we track, the maker's recommended replacement, and other options compared on the Analyzer Index and price. Dates come from each maker's own lifecycle page, last checked Nov 15, 2026; earliest-possible dates that are not yet confirmed are left out.
Add all shutdowns to your calendarCalendar files include reminders 30 and 7 days before each date. Subscribe by URL to receive new retirements as they are announced.
Upcoming (13)
- Claude Sonnet 4.5AnthropicNov 30, 2026 (in 14 days)
- Maker recommends
- Claude Sonnet 5.5Index +24.0 · 0.67× price
- Other options
- Gemini 4 CarbonCheaper, comparable or better · Index +38.2 · 0.33× price
- Gemini 4 ArgonStronger, similar price or less · Index +35.3 · 0.33× price
- Kimi K3.1 ProOpen-weight option · Index +24.0 · 0.67× price
- Gemini 2.5 FlashGoogleDec 1, 2026 (in 15 days)
- Maker recommends
- Gemini 3.5 Flash LiteIndex +4.6 · same price
- Other options
- DeepSeek V4.1 FlashCheaper, comparable or better · Index +38.7 · 0.49× price
- GLM 5.3 FlashStronger, similar price or less · Index +29.9 · 0.56× price
- DeepSeek V4 Flash 0731Open-weight option · Index +24.6 · 0.21× price
- Gemini 2.5 ProGoogleDec 1, 2026 (in 15 days)
- Maker recommends
- Gemini 3.8 FlashIndex +23.9 · 0.44× price
- Other options
- Gemini 3.7 FlashCheaper, comparable or better · Index +25.4 · 0.44× price
- Gemini 4 CarbonStronger, similar price or less · Index +38.6 · 1.2× price
- DeepSeek V4.1 FlashOpen-weight option · Index +21.5 · 0.12× price
- Dec 1, 2026 (in 15 days)
- Maker recommends
- Gemini 3.8 Flash0.44× price
- Gemini Omni 1.1 FlashGoogleDec 3, 2026 (in 17 days)
- Maker recommends
- Gemini Omni 1.2 Flashsame price
- GPT-5OpenAIDec 11, 2026 (in 25 days)
- Maker recommends
- GPT-6.1 SolIndex +24.0 · 0.58× price
- Other options
- Gemini 4 CarbonCheaper, comparable or better · Index +31.0 · 0.58× price
- Gemini 4 ArgonStronger, similar price or less · Index +28.1 · 0.58× price
- Kimi K3.1 ProOpen-weight option · Index +16.8 · 1.2× price
- GPT-5 miniOpenAIDec 11, 2026 (in 25 days)
- Maker recommends
- GPT-6.1 Terra2.9× price
- Other options
- DeepSeek V4 Flash 0731Cheaper, comparable or better · Index +15.3 · 0.26× price
- DeepSeek V4.1 FlashStronger, similar price or less · Index +29.4 · 0.61× price
- GLM 5.3 FlashOpen-weight option · Index +20.7 · 0.69× price
- GPT-5 nanoOpenAIDec 11, 2026 (in 25 days)
- Maker recommends
- GPT-6 LunaIndex +29.7 · 1.5× price
- Other options
- DeepSeek V4 Flash 0731Open-weight option · Index +30.5 · 1.3× price
- GPT-5 ProOpenAIDec 11, 2026 (in 25 days)
- Maker recommends
- GPT-6.1 Sol Pro0.05× price
- o3OpenAIDec 11, 2026 (in 25 days)
- Maker recommends
- GPT-6.1 SolIndex +29.7 · 0.57× price
- Other options
- Gemini 4 CarbonCheaper, comparable or better · Index +36.6 · 0.57× price
- Gemini 4 ArgonStronger, similar price or less · Index +33.8 · 0.57× price
- Kimi K3.1 ProOpen-weight option · Index +22.4 · 1.1× price
- o3 ProOpenAIDec 11, 2026 (in 25 days)
- Maker recommends
- GPT-6.1 Sol Pro0.06× price
- Mar 15, 2027 (in 119 days)
- Maker recommends
- Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image)0.66× price
- Gemini 3.1 Flash LiteGoogleMay 7, 2027 (in 172 days)
- Maker recommends
- Gemini 3.5 Flash LiteIndex -5.6 · 1.5× price
- Other options
- DeepSeek V4 Flash 0731Cheaper, comparable or better · Index +14.4 · 0.32× price
- DeepSeek V4.1 FlashStronger, similar price or less · Index +28.5 · 0.75× price
- GLM 5.3 FlashOpen-weight option · Index +19.7 · 0.84× price
Recently retired
- GPT Image 1OpenAIOct 23, 2026 (in -24 days)
- Maker recommends
- gpt-image-2.5-sunburst or gpt-image-2.5-flare
How to read this
"Index" is the change in Analyzer Index points against the retiring model; "price" is the blended list price (3 input : 1 output tokens) as a multiple of the retiring model's. Other options come from the same rules as each model page's alternatives. Check your own workload in the cost calculator before switching; tokens used per task differ between models, which cost per task shows.