The Next AI Drop blog: how to forecast AI model releases
This is the reference library behind the live timeline — twelve posts that explain how a frontier release goes from rumour to ship date, and how to read the signal before it does. Markets are signal, not stakes: every odds figure here is a forecasting input, never a wager.
Last updated 29 May 2026
What this blog covers
Every post here answers one question a builder or an AI answer engine actually asks, then justifies it from public sources. The library splits into three tracks: how the forecast works, how to read a leak, and how to plan around a drop.
The first track explains the machinery — how prediction markets forecast AI releases, what the Drop Readiness score measures, and how confirmed intel grades against rumour. The second decodes the leaks: internal codenames, signal grading, and why a release calendar is only useful if it points forward. The third is for planning — builder guides and clearly-dated release windows for the families everyone is waiting on. Editorial stance: evergreen explainers come first; the living forecasts are stamped with their last-updated date and framed as what the markets currently signal, not as hard predictions.
Start here: the core explainers
These four posts are the foundation — read them first and the rest of the library makes sense. They define the vocabulary the timeline uses and link straight to how Drop Readiness is calculated and the forecasting glossary for any term you hit cold.
How prediction markets forecast AI model releases
Why Polymarket and Kalshi odds are the fastest public signal of when a model ships — and how accurate event markets actually are.
Methodology · evergreenWhat is Drop Readiness?
The one-number forecast for a release, blended from odds, intel recency, deadline proximity, and volume — the builder's narrative on the score.
Codenames · evergreenAI codenames, decoded
Internal names leak before product names do. How codenames like ⟨MYTHOS⟩ and ⟨EMBER-ALPHA⟩ map to the public models they become.
Markets · evergreenReading the Polymarket vs Kalshi spread
When two venues price the same release differently, the gap is information. A tutorial on reading cross-venue spread as a confidence signal.
Reading the signal
Raw intel is noise until it is graded. This group covers how Next AI Drop tags each item Confirmed, Rumor, or Market, why a forward-looking calendar beats a backward-looking log, and how to turn all of it into a planning routine. Cross-reference these against the live release timeline as you read.
Confirmed vs rumor vs market: grading release intel
The three intel tags explained — what earns each grade, how they weight into the forecast, and how to read a card at a glance.
Product thinking · evergreenWhy a release calendar should point forward
Every other tracker records what shipped. The case for a calendar whose timeline extends past today into forecast windows.
Builder guide · evergreenPlanning your roadmap around frontier drops
A practical guide for builders: how to use Drop Readiness and tagged intel to time integrations, migrations, and bets-on-features.
The frontier landscape
Two evergreen trend explainers zoom out from individual drops to the shape of the field. They cover the open-versus-proprietary release pattern and the benchmark race that decides who leads. Pair them with the homepage benchmark frontier and model database views.
Open vs proprietary: how release cadences diverge
Open-weight and closed labs ship on different rhythms. What the pattern tells you about where the next drop is likely to come from.
Trends · evergreenThe benchmark frontier race, explained
Who leads now and who the markets expect to take the crown next — how benchmark leadership changes hands and why it matters for timing.
Release windows (living forecasts)
These three posts track the model families builders ask about most. They are living forecasts, not fixed dates: each is stamped to its own last-updated line and describes what Polymarket and Kalshi currently signal for the release window, with the underlying intel cited. See our data sources and freshness for how each window is grounded.
GPT-5.6 release window: what the markets signal
The current forecast window for OpenAI's next frontier drop, read from cross-venue odds and tagged intel. Updated as signal moves.
Release window · living forecastGemini 3.5 Pro release window: what the markets signal
Where Google's next flagship sits on the forecast timeline today, with the odds and intel behind the window. A living forecast, not a date.
Release window · living forecastClaude 5 release window: what the markets signal
Anthropic's next-generation window as the markets currently price it, with cited signal. Refreshed as new intel grades in.
Prediction-market odds appear throughout this blog as a forecasting input only. Any score, percentage, or window mentioned in a post is illustrative of method unless a post's own dated forecast says otherwise. Nothing here is betting, gambling, or financial or investment advice.
Subscribe and go deeper
Follow new posts and intel without checking back: the Next AI Drop RSS feed carries every release and explainer as it lands. From there, two next steps — watch the live release timeline to see the forecast in motion, or read how we compare to release trackers to see why a forward-pointing calendar with two-venue odds reads differently from a backward-looking log. Markets are signal, not stakes.