Menu
farexio
Tech

Will Anthropic have the best AI model on the LMArena leaderboard at the end of September 2026?

Resolution: Updated:

In short

The market treats an Anthropic win as highly likely, not certain. The main reason is that an Anthropic model currently holds the top spot, and incumbency on LMArena tends to persist for weeks unless a rival lab ships a new frontier model. A GPT or Gemini release that beats Anthropic's score before the 30 September 2026 snapshot would flip this.

Editorial illustration for: Will Anthropic have the best AI model on the LMArena leaderboard at the end of September 2026?

How the contract works

This contract settles at $1 if the model ranked #1 on the specified LMArena leaderboard belongs to Anthropic when checked on 30 September 2026 at 12:00 PM ET, and at nothing otherwise. Ties are broken first by the underlying Arena score and then alphabetically by company name if scores are identical. A price of, say, 0.30 would mean the market sees roughly a three-in-ten chance of that outcome, purely as a hypothetical unrelated to this specific market. Anyone holding a position can typically sell it before the 30 September settlement date at whatever price the market offers at that moment, rather than waiting for resolution.
What the market thinks happens
$100
Yes90%

The event happens

Costs now
$0.90
If you put in $100
$111
No10%

The event does not happen

Costs now
$0.10
If you put in $100
$1,000

Probability

History starts collecting once the event is tracked

How the price has moved

The only pricing data available comes from a single venue, Polymarket, currently showing 90% on $64,230 of recorded volume; there is no second venue to compare against and no reported intraday or weekly move to describe. A consensus this high with volume this modest suggests a market that has settled into a status-quo view โ€” Anthropic already leads the named leaderboard โ€” without much active disagreement or repricing pressure so far. The absence of a documented price history here is itself informative: it points to a market that has not yet faced a catalyst, such as a rival model launch, significant enough to force a visible repricing.

Analysis

Context

LMArena is a crowdsourced benchmark where users compare two anonymous AI model outputs side by side and vote for the better one; those votes feed an Elo-style score used to rank every major model from OpenAI, Google, Anthropic, Meta and others. The specific board this market tracks is the Text Arena Overall leaderboard with style control turned off, meaning answers are judged as users see them, including formatting and verbosity, rather than adjusted to strip out stylistic advantages. The three labs that have traded the top spot over the past two years are Anthropic (Claude), OpenAI (GPT) and Google (Gemini), each releasing new flagship models every few months in a race that regularly reshuffles the board.
The consensus price across tracked venues sits at 90%, with all $64,230 in recorded volume concentrated on a single venue, Polymarket. A single-venue market means there is no cross-venue spread to read for disagreement between traders in different pools โ€” the 90% figure is the only signal available, and it should be read as one market's aggregated view rather than a convergence of several independent ones. A reading this high generally reflects a status-quo bet: Anthropic already occupies #1 on the named leaderboard, and the market is pricing the chance that this holds for roughly five more weeks rather than the chance of Anthropic reaching the top from behind. That is a meaningfully easier condition to satisfy than starting from zero, which is the main reason the price sits close to full confidence rather than near a coin flip. The volume, just over $64,000, is modest for a Polymarket contract, which means the 90% figure can move on relatively small trades if a competing lab announces a new model release; a market this thin does not need much capital to reprice. The structural tie-break rule adds a small extra cushion for Anthropic: if its score ends up exactly matched with a rival, alphabetical order settles the tie, and "Anthropic" precedes both "Google" and "OpenAI" alphabetically. That detail is minor compared to the underlying incumbency, but it is one of the few mechanical facts in this market's own rules that pushes in Anthropic's favor rather than against it.

What moves the probability

  1. Current #1 position

    An Anthropic model already sits atop the named LMArena leaderboard, which is the single biggest reason the price is high. Leaderboard leads on LMArena have historically persisted for weeks between major model launches, so simply holding the position through 30 September 2026 without a new challenger is enough for a Yes.

  2. Competing lab release cadence

    OpenAI and Google have each shipped new flagship models multiple times in the past two years, and either could release a model before the snapshot that outscores Anthropic's current entry. This is the largest source of downside risk to the price, since a single high-profile launch in September could reorder the board quickly.

  3. Style control turned off

    With style control off, the ranking reflects raw user preference including formatting and verbosity rather than a stripped-down accuracy comparison. This setting has, in past leaderboard cycles, sometimes favored models with more elaborate or structured output styles, which can work for or against whichever lab currently leads depending on its house style.

  4. Alphabetical tie-break

    If Arena scores end in a tie, the rule falls back to alphabetical order by company name, and Anthropic precedes Google and OpenAI. This is a small structural nudge in Anthropic's favor that only matters in the narrow case of an exact score tie.

  5. Single snapshot timing

    Resolution depends on one specific check at 12:00 PM ET on 30 September 2026, not an average over time. A leaderboard update, temporary model listing, or a fresh release landing just before that moment could change the outcome even if Anthropic led for most of the month.

The case for

  • Anthropic's model remains ranked #1 on the LMArena Text Arena Overall leaderboard through the 30 September 2026 checkpoint without being displaced.
  • No rival lab ships a new model before that date that scores higher under the style-control-off methodology.
  • If scores end in an exact tie, the alphabetical tie-break rule resolves in Anthropic's favor over Google or OpenAI.
  • The leaderboard snapshot taken at 12:00 PM ET on 30 September 2026 captures Anthropic still on top rather than a transient reshuffle.

The case against

  • OpenAI or Google releases a new flagship model before 30 September 2026 that overtakes Anthropic's current score.
  • LMArena's crowdsourced voting produces enough score movement over five weeks to shift the order even without a new release.
  • The resolution source becomes unavailable at check time and the fallback snapshot captures a different leaderboard state than expected.
  • A newer, unranked entrant model climbs quickly enough on user votes to surpass Anthropic before the deadline.

What to watch

The key date is 30 September 2026 at 12:00 PM ET, when the arena.ai Text Arena Overall leaderboard with style control off is checked for the #1 model. Between now and then, any announced or released flagship model from OpenAI, Google, Meta or another lab is the main event that could alter the ranking, since LMArena updates its board continuously as new models are added and as user votes accumulate. Readers should also watch for any changes to the leaderboard's methodology or availability, since the settlement rules specify a fallback to the next available snapshot if the source is down at check time.

Trade this contract

Venues (1)

Open on PolymarketYes 0.90
  • gas covered

More about this event

Venues (1)

Probability

  • Will Anthropic have the best AI model at the end of September 2026?90%
  • Will OpenAI have the best AI model at the end of September 2026?6%
  • Will Google have the best AI model at the end of September 2026?3%
  • Will SpaceXAI have the best AI model at the end of September 2026?2%
  • Will DeepSeek have the best AI model at the end of September 2026?0%
  • Will Alibaba have the best AI model at the end of September 2026?0%
  • Will Xiaomi have the best AI model at the end of September 2026?0%
  • Will Microsoft have the best AI model at the end of September 2026?0%

Resolution rules

Determined by
https://arena.ai/leaderboard/text/overall-no-style-control
Resolution date

Resolution is based on the arena.ai Text Arena (Overall) leaderboard with style control off, checked on 30 September 2026 at 12:00 PM ET. The market resolves Yes if the #1-ranked model at that moment belongs to Anthropic, with ties broken first by the underlying Arena score and then alphabetically by company name. If the leaderboard is unavailable at check time, the next available snapshot is used per Polymarket's standard rules; all tracked venues here rely on this same single source, so there is no cross-source discrepancy to account for.

Calculation methodology โ†’

Local context

English-language tech and finance audiences follow the Anthropic-OpenAI-Google leaderboard race closely because it doubles as a proxy for competitive position among companies tied to major public markets โ€” Alphabet directly, and OpenAI indirectly through Microsoft's investment and commercial partnerships. A shift in LMArena's #1 spot is often read by this audience as an early signal of which lab's technology is gaining developer and enterprise mindshare, ahead of quarterly earnings commentary from Alphabet or Microsoft that references AI product traction.

Common questions

What exactly settles this market and when?
The arena.ai Text Arena Overall leaderboard with style control off, checked specifically on 30 September 2026 at 12:00 PM ET. The market resolves Yes if the model ranked #1 at that moment is owned by Anthropic.
What does the current market price mean?
The price is the market's collective estimate of the probability that Anthropic holds #1 at the deadline, not a guarantee. A price near 90% reflects a market that sees this as likely but not settled, since a new competing model release could still change the outcome before 30 September 2026.
What happens if the leaderboard is down or the result is ambiguous at check time?
The settlement rules specify that if the resolution source is unavailable at the check time, the market uses the next available snapshot according to Polymarket's standard rules. This avoids the market being stuck if arena.ai has downtime exactly at the deadline.
How are ties between labs broken on the leaderboard?
Ties are broken first by the underlying Arena score behind the displayed ranking, and if that is also tied, alphabetically by company name. Since "Anthropic" comes before "Google" and "OpenAI" alphabetically, an exact tie would resolve in Anthropic's favor.
Why does it matter that style control is turned off?
With style control off, the ranking reflects raw human preference for an answer's presentation as well as its content, rather than adjusting to isolate pure accuracy or reasoning quality. This setting can favor models whose typical output style โ€” length, formatting, structure โ€” tends to appeal to evaluators, regardless of which lab built them.
Has Anthropic held the #1 spot before, and does that matter here?
The market's high price implies Anthropic currently holds #1 on this specific leaderboard, and leaderboard leadership on LMArena has historically tended to persist between major model releases rather than flip every few days. That pattern is one reason the market leans toward Yes, though it is not a guarantee against a new release displacing the lead before the deadline.

Related prediction events

Tokenized stocks

Market-implied probabilities that provide context for this assetโ€™s catalysts.

Related events

90%/ 10%
Yes / No
Open on Polymarket