As of the last-updated date above, prediction markets signal an imminent-to-expected Claude 5 window: a near-term arrival that markets favour without committing to a hard date. This post explains what that signal is made of and where it points — the live forecast card on the home timeline carries the current numbers, because this reading moves hourly.
We read Polymarket and Kalshi as a forecast of when a model ships, never as something to bet on. Nothing here is betting, gambling, or investment advice — the odds are a timing instrument, and every number in this post is illustrative unless you are looking at the live card.
What markets currently signal
Markets currently lean toward a near-term Claude 5 arrival, placing it in the imminent-to-expected band rather than the rumoured tier. That reading comes from where Polymarket and Kalshi sit on the question of whether Anthropic ships its next flagship within the active forecast window, combined with how recently fresh intel has landed. The ⟨MYTHOS⟩ codename anchors the watch: when a sighting attaches to a known codename, the signal stops being noise and becomes a datable lead.
Treat the exact figures as a moving target. A representative snapshot might show a Drop Readiness score in the high-70s with the stage marked imminent — but that is an example, not a quote. For the authoritative call, read the live Claude 5 forecast card, which recomputes the moment new odds or intel arrive.
Is Claude 5 out yet?
No — as of the date shown above, Claude 5 has not shipped. That status is the single most hourly-sensitive fact on this page: a launch can land between one refresh and the next, and a card can flip from imminent to released without warning. We do not freeze a guess into this prose for that reason.
If you want a status you can trust to the hour, check the live release timeline rather than this post. The timeline reflects the current state of the markets and the intel feed; this article explains the method that produces it.
How we read the Claude 5 window
We score the window with one formula, applied identically to every model: DR = .45 odds + .25 intel + .20 deadline + .10 volume. Odds are the blended Polymarket and Kalshi probability that Claude 5 ships inside the active window; intel is the recency and grade of fresh signal; deadline rewards proximity to a credible window edge; volume weights how much market conviction sits behind the odds. The result lands the model in one of three stages — imminent, expected, or rumoured — and refreshes hourly. Read the full computation in our methodology.
Claude 5 is also a case where the two venues do not always agree. As an illustrative example: if Polymarket reads the ship probability at 82 and Kalshi reads it softer at 76, that six-point gap is itself a signal — venues disagreeing on timing usually means the market is still pricing in fresh, uneven information. We treat the wider of the two as the optimistic edge of the window and the narrower as the conservative edge. Those numbers are made up to show the shape of the read, not a live quote. For how to interpret a divergence like this, see the Claude 5 cross-venue spread.
The intel behind the window
The intel half of the Claude 5 read is built only from publicly observable signal, and each piece carries a grade. A ⟨MYTHOS⟩ entry appearing in a public model-catalog listing, a documentation schema gaining an undocumented field, or an official changelog note are the kinds of things that move the intel weight. We tag each one Confirmed, Rumor, or Market so the score reflects evidence quality, not volume of chatter.
Crucially, an early catalog sighting is graded as a rumour or a market move, not as a fact that Claude 5 is live — a visible endpoint is a lead to verify, not a launch. That discipline is what keeps the score honest when a codename surfaces before any announcement. For how each grade is defined and why it matters, read how we grade Confirmed vs Rumor vs Market signal, and see the data behind this forecast for what we draw on and what we do not.
Track Claude 5 live
This page is a snapshot of method; the timeline is the live instrument. Everything above explains how the Claude 5 window is constructed, but the window itself shifts every hour as odds reprice and intel lands — so bookmark the live Claude 5 forecast card and check it against this read whenever you are planning around the drop. When the markets move, the card moves first.
Watching more than one flagship? The same method tracks the next OpenAI release — see the GPT-5.6 window for the parallel forecast, and start from the Drop Readiness score if you want the story behind the number.
Next AI Drop is independent and not affiliated with Anthropic, Polymarket, or Kalshi. We build on Polymarket and Kalshi as forecast sources and add the second venue, hourly graded intel, and a single Drop Readiness score. Next AI Drop is operated as a solo project from Amsterdam, Netherlands — contact hello@nextaidrop.com.