01A vintage records one growing season, so the same wine can sit in a different window each year.
Takeaway 2
02Cellared calculates an estimated window for each bottle, and the useful output is how the windows relate to one another.
Takeaway 3
03Heat exposure and bottle variation move the real range, so adjust by what you know about each bottle.
Takeaway 4
04Sort vintages into open now, open soon with a plan, or hold for a stated reason.
To compare wine vintages in your own cellar, line up the same wine across years, read each bottle's estimated drinking window side by side, and decide which vintage to open now and which to hold. The comparison works because a vintage is a record of one growing season, and each season ripens a wine differently. With Cellared connected to ChatGPT, you can ask for the comparison in plain language, such as "Compare the calculated drinking windows Cellared returns for my Cabernet vintages," and get an answer built from the bottles you own.
This guide explains what a vintage comparison tells you, what it cannot, and how to turn the result into a decision.
Why vintages differ
A vintage is the year the grapes were harvested. Weather shapes the fruit that year: the amount of rain, the heat of the summer, the timing of harvest. The Wine Institute's report on the 2018 California harvest, for example, describes a long growing season with moderate spring and summer temperatures and a harvest that began ten days to three weeks later than the year before. Napa Valley Vintners describes the same season as long, cool and even, with extended fall sunshine. A season like that tends to produce wines that develop more slowly, which is one reason a vintage can sit in a different window than its neighbors.
City AM's wine columnist makes a related point: many strong vintages do not receive the highest scores because they take longer to become approachable, and a year that critics call phenomenal may mean the wine is drinking beautifully now. In other words, a vintage's reputation and its readiness are separate questions. A comparison across your own bottles keeps both in view.
Producers matter as much as the weather. Two estates in the same valley and the same year can make wines with different structure and oak, which means different windows. Compare the same producer and wine across vintages when you can. That is a vertical, and it is the cleanest test. The guide to building a Bordeaux vertical shows how.
What a calculated window gives you
Cellared calculates a drinking window for each bottle using the Cellared Ageability Index (CAI), described on the methodology page. The window is an estimate. It places a bottle in a phase, such as approaching, at peak, or past, and it gives a range of years. Wine Spectator describes windows the same way: a range of likely best drinking, drawn from experience with that type of wine, that you can adjust for your own taste.
When you compare vintages, the useful output is the relationship between windows. If the 2018 sits at peak and the 2020 is still approaching, you have an answer for tonight and a plan for later. If both sit at peak, the question becomes which one you want more. If one has passed its window, move it up the queue.
Free on the App Store
Keep your cellar and timing guidance together
Save unlimited bottles with a free Cellared account. Drinking-window guidance is available when sufficient bottle and vintage information exists.
Ask for a comparison of a specific wine or a grape you hold in several years. Examples:
Compare the calculated drinking windows Cellared returns for my Cabernet vintages.
I own the 2016, 2018 and 2019 of the same Barbaresco. Which should I open this year, and which should I hold?
Show my Bandol Rouge vintages in order of how soon they reach their windows.
A good answer lists each vintage you own, its calculated window and its phase. It should present windows as estimates. If a vintage is missing, check whether the bottle is recorded under a slightly different name or region. If you grant write permission, you can correct a bottle's region or quantity from the same conversation, and the next comparison will use the cleaner record.
You can also ask for reasoning: "Why does the 2020 have a later window than the 2018?" The answer may refer to the wine's structure, the vintage conditions in the data Cellared holds, and the inputs behind the CAI. Treat that reasoning as a starting point for your own tasting. The Cellared personalization guide describes how windows adjust to a collector's preferences over time.
Reading the comparison with care
Windows are ranges. A bottle two years from the end of its window is not on a cliff. Wine Spectator's columnist notes that wines past their windows are generally safe to drink, though some may be unpleasant, and that older wines can be an acquired taste. Your own preference decides how much weight to give the late end of a range.
Storage moves the range. City AM notes that heat accelerates maturity. If one vintage lived in a warmer spot, its real window may be earlier than the calculated one. Adjust by your own knowledge.
Bottle variation exists. Even with perfect storage, two bottles of the same vintage can differ. The practical remedy is the oldest piece of cellar advice: when you hold several bottles, open one and taste it. Record your note. That tasting is the most reliable update to your sense of the whole case.
Early is not always wrong. Some collectors enjoy a structured wine young, with decanting, and some wait for the savory notes of age. Neither is a mistake. The comparison gives you the map, and your palate chooses the route.
From comparison to decision
Once you have the windows side by side, sort the vintages into three actions.
Open now. Choose bottles at peak or near the end of their range, especially if storage was imperfect. These are the bottles that gain nothing by waiting.
Open soon, with a plan. Choose bottles entering their windows. Mark them for the next season, a holiday or a dinner you already have in mind.
Hold deliberately. Choose bottles still approaching their windows, and give each a reason: a milestone, a vertical you are building, or a producer you trust to reward patience. A bottle held without a reason tends to be forgotten, and the cellar triage guide is built for that situation.
Tell ChatGPT what you decided. With write permission, ask it to record the opening when you pour, and save a 5-point rating with a note. The next comparison, a year from now, will include what you learned.
A worked example, in prose
Imagine a collector with three vintages of the same Napa Cabernet. The 2016 is mature, the 2018 is entering its range, and the 2021 is young. Asked to compare the calculated windows, a good answer sorts them in that order and flags the 2016 as the bottle to open soon. It suggests the 2018 for a dinner this season and holds the 2021, with a note on when its window is estimated to begin. The collector disagrees about the 2018, because that case lived near a warm wall in the garage, and moves it earlier. That final adjustment is the useful part. The connection supplies a structured estimate, and the collector supplies the knowledge of the bottle.
Confirm the bottle records. A comparison is as good as the inventory behind it. Check vintage, producer and region on each bottle, because a mistaken year changes the window. Confirm the permissions you granted, so you know whether ChatGPT can fix a record. And keep the claim modest: the result is an estimate calculated from the bottle, not a guarantee. The Cellared in ChatGPT page lists what the connection reads and what it can change with your permission.
Keep your saved bottles and available drinking-window guidance together. Free Cellared accounts support unlimited saved bottles; estimates require sufficient bottle and vintage information.