Kalshi's market on which company has the top AI model before 2027 lists 12 companies. The company that actually holds first place on the leaderboard the contract settles on, Anthropic, is not one of them. That is the strange fact this whole board sits on, and it is why our eight-model panel came back below the market on all but one of the prices.
The Quick Answer
Nobody qualifies today, and the panel thinks nobody is likely to. Anthropic holds every share of first place on the settlement leaderboard and has no contract, so every price on this board is a bet on one of 12 challengers dislodging it before January 1, 2027.
Among the names you can trade, the panel's best value is Meta at 16% against a 13.5-cent price. OpenAI is the market favorite at 33.5 cents. The panel has it at 12.3%.
Prices are Kalshi midpoints as of August 22, 2026. The panel never saw them.
Stay ahead of the markets.
Daily insights and expert picks on Kalshi, Polymarket, and what's moving markets.
Free forever. Unsubscribe anytime.
Why Nobody Qualifies Today
Every contract carries the same one-line summary: a company wins if it "has a #1 ranked AI model before Jan 1, 2027." The binding terms are stricter in three ways. The measure is the upper-bound rank (UB) on LMArena's text leaderboard with the "Remove Style Control" toggle applied, a view you have to switch on deliberately, and in that column models too close to separate share a rank. The check happens at 10:00 AM ET every day, and a company needs to hold the top on only one of them. Then the sentence that undoes all of it, word for word: "If they have an LLM that is tied with another LLM, then the Payout Criterion is not fulfilled." A model has to be alone at the top.
On that settlement view, as of the leaderboard's August 21 vote cutoff, five models share the number-one rank and all five are Anthropic's. Their raw scores differ slightly, but none is provably better than the others, which is exactly the tie the contract describes. Read strictly, no company on the planet qualifies right now, including the one winning.
The Board
Twelve companies, the market's price for each, the panel's number, and where each company's best model actually sits in the column that settles the contract.
| Company | Best model on the settlement view | UB rank | Market | Panel |
|---|---|---|---|---|
| OpenAI | gpt-5.5-high (row 22) | 14 | 33.5¢ | 12.3% |
| xAI | grok-4.5 (row 37) | 30 | 20.5¢ | 4.3% |
| Meta | muse-spark-1.2 xHigh (row 8) | 3 | 13.5¢ | 16.0% |
| Moonshot AI | kimi-k3-max (row 17) | 7 | 13¢ | 6.0% |
| Z.ai | glm-5.3-max (row 15) | 6 | 11¢ | 6.5% |
| Alibaba | qwen3.8-max (row 11) | 6 | 10¢ | 8.8% |
| Nvidia | nemotron-3-ultra-550b (row 51) | 33 | 8.5¢ | 1.1% |
| Deepseek | deepseek-v4-pro (row 40) | 32 | 8¢ | 3.8% |
| ByteDance | dola-seed-2.0-pro (row 43) | 34 | 7.5¢ | 1.5% |
| Baidu | ernie-5.1 (row 24) | 15 | 6.5¢ | 2.8% |
| Mistral | mistral-large-3 (row 84) | 71 | 4.5¢ | 0.9% |
| 01A1 | no model listed under this organization | n/a | 4¢ | 0.5% |
Market = the midpoint of Kalshi's live bid and ask, fetched August 22, 2026; several of these books are wide, so the spread on screen, not the midpoint, is what a position actually costs. Panel = the median of eight independent model estimates after a revision round. Rows and UB ranks come from the settlement-view leaderboard, vote cutoff August 21, 2026. Kalshi's contract still says "xAI"; the leaderboard files Grok under SpaceXAI, the name the company took after SpaceX acquired xAI in an all-stock deal on February 2, 2026. Kalshi also lists a "Zhipu AI" contract at zero that has never traded; Zhipu rebranded as Z.ai in July 2025, so the board carries two contracts for one company.
Set the middle columns side by side and the market's shape stops making sense. The three companies closest to the top in the column the contract names, Meta, Alibaba and Z.ai, all trade under 14 cents. The two priced highest, OpenAI and xAI, sit 14th and 30th in that column. Traders have this board close to backwards from the leaderboard it settles on.
There is one more way to read the two price columns. Kalshi flags the board as not mutually exclusive, so each column can be added up and read as the number of companies expected to reach a sole number one before 2027. The market's prices sum to 140.5 cents, or about one and a half qualifiers. The panel's sum to 64.5, about two thirds of one. That gap is the whole disagreement.
Why The Panel Won't Pay 33.5 Cents For OpenAI
OpenAI's best model on the settlement view, gpt-5.5-high, sits 14th in the column that counts, with a rating that 59,545 votes have pinned firmly in place. Whatever OpenAI does on this board, it does with a new model. The panel still gave it the second-highest number, because it has the strongest record here of shipping a release that jumps a long way at once, and it puts everything on this leaderboard, 44 models, more than any other company. The 12.3% is a forecast about OpenAI's next model.
Shipping is the easy part. The contract pays only for a release that stands alone above five Anthropic models, which is a much narrower event. The GPT seat, which had priced OpenAI at 19 in the first round and cut it to 13 after reading the others, put the board in one line.
"The board is shaped by strict sole-UB-1 settlement risk, with Meta closest today and OpenAI retaining the largest latent release-jump upside." — GPT
Kalshi runs a second top-AI market that asks who is on top on December 31 alone, and we priced that board separately. There, Claude trades at 69 cents and ChatGPT at 13. OpenAI is more than two and a half times as expensive on the board where Anthropic has no contract as it is on the board where Anthropic does.
Why Meta Is The One Name Priced Too Low
The reason is a toggle. On the settlement view, muse-spark-1.2 (xHigh) sits third in the UB column on just 3,257 votes. Flip style control back on, which is the leaderboard's default public view and not the one that settles this contract, and the same model is tied for first with three Anthropic models. Meta is already number one by the ranking Kalshi told you not to use.
A tie pays nothing, and the panel priced it that way. What the toggle shows is how narrow the gap is: one release stands between Meta and the top, and Meta ships fast, with 21 models on the leaderboard and a muse-spark line that has gone from muse-spark to 1.1 to 1.2 in short order. The market's 13.5 cents implies about a 1-in-7 chance. The panel's 16% is closer to 1-in-6.
The panel's clearest single trigger is the next muse-spark landing clearly above the Anthropic pack with enough votes to prove it. A qualifying day would close Meta's contract and prove the tie clause can be beaten.
Can The Leaderboard Be Gamed?
The settlement source has a published record against it. The Leaderboard Illusion, from a team spanning Cohere Labs, AI2, Princeton, Stanford, Waterloo and the University of Washington, argues the arena is distorted by undisclosed private testing: large labs test many variants and publish only the best. It counts 27 private variants Meta tested before the Llama-4 release. A second paper, Improving Your Model Ranking on Chatbot Arena by Vote Rigging, concludes that rigging only hundreds of new votes can move a target model's rank.
Neither paper accuses this board of being manipulated, and neither changes what the contract says. They matter because the ratings behind the two best-placed challengers, Meta and Z.ai, each rest on a few thousand votes, while OpenAI's rests on 60 thousand. A rating built on three thousand votes moves for reasons a rating built on 60 thousand does not, and the contract pays on a single morning's reading.
The Meta finding cuts both ways, which is why the panel did not treat it as a reason to sell. A lab willing to test 27 variants to find its best one is optimizing hard for exactly this leaderboard.
Where The Panel Changed Its Mind
After round one each seat read the other seven anonymously and could revise. Several had built a bull case on the debut window: a brand-new model with few votes carries a wide confidence interval, which might briefly push its upper bound above everyone else's and win a morning. Opus took that apart.
"Sole rank(UB) 1 requires your lower bound above everyone's upper bound, so wide-CI debuts can't win and multi-variant labs tie themselves." — Opus
In plain terms, few votes make a model look potentially great and potentially terrible at the same time. That gets it into the tie at the top, which pays nothing, and it stops the model from clearing that tie. The winning shape is the opposite: a big rating jump plus enough votes to prove it. A lab that submits many models is partly working against itself, because its own entries crowd the top band. Anthropic's top two models sit 0.2 Elo points apart, which is exactly why nobody qualifies today.
The other seats moved toward that reading. GPT cut Meta from 31 to 17, the largest single revision of the run. Gemini cut OpenAI from 25 to 11.5 after being the most bullish seat on it in round one.
Kimi went the other way, raising OpenAI from 7 to 15 on the argument that step-change releases happen every quarter and there are a lot of mornings left to catch one. The spread across seats on OpenAI still fell from 18 points to 7, more convergence than this panel usually produces.
Who Else Has A Case?
Alibaba, 10¢, panel 8.8%. qwen3.8-max sits sixth in the UB column, eight places closer to the top than OpenAI's best, and Alibaba has 42 models on the leaderboard, second only to OpenAI. The GPT seat came closest to meeting the market, at 9.
Z.ai, 11¢, panel 6.5%. glm-5.3-max is also sixth, on only 3,751 votes. The GLM seat, pricing its own maker's contract, was among the more cautious voices. The book is 2 cents bid against 20 asked, which is barely a quote.
Moonshot AI, 13¢, panel 6.0%. kimi-k3-max sits seventh on 15,054 votes, a firmer position than Z.ai's, but Moonshot has only 7 models listed, fewer shots on goal than the labs around it.
The panel wrote off the rest. Nvidia's best model is 33rd in the UB column and Mistral's is 71st. ByteDance has exactly one model on the entire leaderboard, at 34th. No seat priced any of the three above 2%. The DeepSeek seat, pricing its own maker's contract at 2.5%, was the most bearish voice about the board as a whole.
"The tie clause requires sole UB=1 separation above five clustered Anthropic models, making this a clean-step-change contract where only Meta's proximity and OpenAI's release history justify >5% probabilities." — DeepSeek
By that standard, ten of the 12 contracts trade above the highest number any seat gave them. Only Meta and Alibaba have a seat willing to meet the market.
Why Anthropic And Google Are Not Listed
Anthropic holds all five first-place slots on the settlement view, led by claude-opus-5-high. Google's gemini-3.7-flash-high is the best-placed model from any other company, tied third. Neither has a contract on this board.
Kalshi does not publish its listing decisions, so we cannot tell you why. We can tell you the exchange priced the same fact elsewhere. On its weekly top-model market for the week of August 24, claude-opus-5-high was quoted at 91 bid against 94 asked. Traders on that board were near-certain who is winning. Traders on this one were paying 33.5 cents for the company sitting 14th. Eleven of these contracts opened on January 1, and not one has had a qualifying day since.
The Bottom Line
The market page describes a contract on who builds the strongest AI, with a favorite you can name and a lot of mornings for it to happen. The terms describe one model standing statistically alone above every other on a 394-row leaderboard, on a specific morning, measured by a column most people never look at. The second is much harder, and on 11 of these 12 contracts the price pays for the first while the second settles.
The one name left standing is Meta at 13.5 cents, three places from the top and already tied for first under the other toggle. It is a small edge on a thin book, and the spread on screen is what it costs. Prices and leaderboard positions are as of August 22, 2026; this page is re-scored when the story moves, and every number here gets graded in public on the full graded scoreboard.
Prices on this page are Kalshi's book. If you also trade on Polymarket, code OS4 gets new users a $50 trading bonus on a $10 deposit — affiliate link; terms as stated by Polymarket; 18+, availability varies by state.
More on this: Best AI Model 2026: Who Finishes The Year On Top · Anthropic IPO Odds: The Leader You Cannot Trade Here · What Counts As An OpenAI IPO On Kalshi · The Panel's Full Graded Scoreboard
