Frequently Asked Questions

Why use prediction markets if this isn't a betting site?

Prediction markets are the fastest public signal of when a model actually ships, so we read their odds as a probability rather than a wager. When informed forecasters put money behind a date, the resulting price reacts to real information — a leaked SKU, a docs change, an exec hint — often before any official post.

We take that price as an input and never as an invitation to trade. The number tells you how likely a release is by a window; what you do with that is planning, not betting. See the markets disclaimer for how we draw that line.

Is Next AI Drop a betting or gambling site?

No. Next AI Drop takes no bets, holds no funds, runs no market, and has no affiliation with any venue. We read public odds and present them as forecast signal next to intel and a Drop Readiness score.

There is nothing here to stake and nothing to deposit. The product surfaces what markets already say in public; it does not let you act on it. The full non-affiliation and no-wager position lives in the markets disclaimer.

Is this financial or investment advice?

No. Nothing on Next AI Drop is financial, investment, trading, or betting advice. The odds, scores, and forecast windows are informational signals to help you plan around releases, and every decision you make from them is your own.

We publish method and ranges, not recommendations. The scope of that disclaimer and the governing terms are set out in the terms of use.

How is Drop Readiness calculated?

Drop Readiness is a single 0–100 score per release, computed as DR = .45 odds + .25 intel + .20 deadline + .10 volume, refreshed hourly. Forty-five percent is blended Polymarket and Kalshi odds, twenty-five is intel recency and strength, twenty is how close the expected window is, and ten is market volume.

Higher means more imminent on current evidence. Releases group into three stages — imminent, expected, and rumoured — as the score rises and the window tightens. The full computation, including how each input is normalised, lives in the full methodology.

What counts as intel?

Intel is the public evidence that a release is coming: official changelogs, public model-catalog listings, indexed public posts, and codename mentions. Each item is source-linked and tagged Confirmed, Rumor, or Market so you can weigh it.

Confirmed is on-record from the provider, Rumor is unconfirmed public chatter, and Market is a move in the odds. The tag and the link travel with every item, so a strong-looking claim and a thin one never read the same. How we source and verify intel is detailed in our sources and trust policy.

What is a codename?

A codename is a provider's internal or pre-release name for a model that surfaces in public before the official name does. We map each one to the public model it most likely becomes, with a confidence read on the match.

Codenames render bracketed and uppercase in monospace across the site — for example ⟨MYTHOS⟩, ⟨GOLDFISH⟩, or ⟨EMBER-ALPHA⟩. Live mappings sit on the codename watch, and the term itself is defined in the glossary of terms.

What does Next AI Drop cover — models, IDEs, agents?

Next AI Drop covers frontier models plus the developer stack around them: IDEs, coding agents, and dev infrastructure such as Cursor, Composer, and Devin-class agents. The categories span text, image, video, audio, code, and agents.

If a drop changes how builders ship, it belongs on the timeline, and the scope widens as the ecosystem does. The full set of tracked releases and their specs is on the model database.

How often is it updated?

Next AI Drop refreshes hourly, with a visible last-updated timestamp on the live page. That cadence is the point of difference, not a detail.

Dedicated editorial release trackers publish on a multi-day or weekly rhythm, so a codename surfacing or an odds move can sit unindexed elsewhere for days while it appears here within the hour. How that hourly intel is gathered and labelled is described in our sources and trust policy.

Is it free?

Yes. Next AI Drop is free, with no account, no paywall, no ads, and no paid placement. Every release is ranked on the same evidence, so no provider can buy a position on the timeline or the forecast board.

It is a solo project run from Amsterdam and kept free by design. More on who is behind it is on the about page.

Where do the odds and data come from, and are sources cited?

Odds come from two prediction markets only — Polymarket and Kalshi — and we name no other odds source. Intel items each carry a link to the underlying public post, changelog, or listing.

Every claim is therefore checkable at its origin, paired with a Confirmed, Rumor, or Market tag rather than presented as bare assertion. The sourcing rules and what we will and will not cite are set out in our sources and trust policy.

How accurate are the forecasts?

Forecasts are probabilistic, not guarantees. A high Drop Readiness score means a release looks likely on current evidence, not that it is certain — odds and scores move as new signal arrives, and announced windows can slip.

Treat the number as a calibrated probability you keep updating, not a fixed date. Any DR value, odds figure, or window shown in static pages is illustrative; live numbers live on the home timeline. How the score is built sits in the full methodology, and its limits in the markets disclaimer.

What is the cross-venue spread?

The cross-venue spread is the gap between Polymarket and Kalshi odds on the same release, surfaced as its own signal. The two venues attract different participants and resolve questions differently, so the gap carries information of its own.

A widening spread can flag uncertainty or fresh information before either side fully reprices — which is why we show it rather than average it away. The role of the spread in the score is covered in the full methodology.

How do I subscribe or get updates?

Subscribe to the RSS feed at /api/feed to get new and updated forecasts in any reader. There is no email list and no account to create.

For a human-readable record of notable product and data changes, follow the product changelog.

How do I report a correction?

Email hello@nextaidrop.com to report an error or a mislabelled item, and it gets corrected when it is wrong. Notable changes are logged publicly.

Source links exist precisely so issues are easy to pinpoint — include the link and what looks off, and the fix shows up in the product changelog. Our standard for what we publish is in our sources and trust policy.