Which company has the best AI model end of September?
Anthropic is still the likeliest winner because Claude Fable 5.1 appears to be leading the relevant leaderboard right now, but the lead does not look unassailable. OpenAI’s GPT-6 Astra is close enough that a late-month swap is a meaningful risk, so I would price Yes high but below the market’s near-certain level.
Analysis
Anthropic has the clearest current case because the market is not asking for a vague general notion of superiority; it is asking who sits in first place on a specific arena leaderboard at a specific check time. On the evidence provided, Claude Fable 5.1 is currently first on the relevant overall text arena and also leads a separate quality-oriented leaderboard. That combination matters because a leader that is already on top of the exact type of ranking used for resolution has a strong structural advantage going into the final day of September.
The main counterweight is that the margin does not appear large. OpenAI’s GPT-6 Astra is described as close, sometimes effectively tied, and in some task-specific settings potentially better. That means the outcome is less about Anthropic being clearly superior in a broad sense and more about whether the exact leaderboard scoring and any late updates preserve a narrow lead. When the gap is small, even modest shifts in model releases, evaluation refreshes, or leaderboard recalculations can change the order at the end.
Market sentiment also strongly favors Anthropic, with quoted probability levels in the mid-90s, and very high traded volume suggests the market has already absorbed a lot of information. That said, prediction markets can overstate certainty when a favorite is widely accepted early and the remaining uncertainty is concentrated in a small set of close competitors. The fact that several independent summaries describe Fable 5.1 and Astra as near peers implies this is not a locked event, just a heavily Anthropic-favored one.
The biggest practical uncertainty is timing and leaderboard dynamics. The resolution depends on a check on September 30 at noon ET, not on a final month-end average, and the market also depends on the specific arena table and ranking details remaining available. If Anthropic holds first place through that check, the Yes outcome is easy; if OpenAI posts a late improvement or the scoring mechanics shift slightly, the result can flip despite Anthropic remaining among the top tier overall.
Arguments
For
- Arguments for Yes: Anthropic currently appears to own the top spot on the specific ranking system that will determine the market.
- Arguments for Yes: Recent independent summaries still place Claude Fable 5.1 at or near the top across general quality and agentic task evaluations.
Against
- Arguments against Yes: OpenAI’s GPT-6 Astra is close enough that a small leaderboard movement could overtake Anthropic before resolution.
- Arguments against Yes: The market is sensitive to the exact resolution mechanism, and narrow ties or metric-specific differences could produce a non-Anthropic winner.
Key drivers
- Claude Fable 5.1 is currently reported as first on the relevant arena leaderboard, which directly supports the Yes outcome.
- GPT-6 Astra is close enough to create genuine late-month reversal risk if the leaderboard updates or ties are broken differently.
Risk factors
- A late OpenAI improvement could move GPT-6 Astra into first place by the September 30 check.
- Any change in leaderboard availability, scoring granularity, or tie-breaking could make the final order less favorable to Anthropic.
Scenarios
Best case
Claude Fable 5.1 remains first on the arena leaderboard through the September 30 check, with no late update or tie-breaker issue, so Anthropic cleanly wins the market.
Most likely
Anthropic holds a narrow but real lead into the final check, making Yes the most likely outcome, though not by enough to justify near certainty because OpenAI remains close and capable of a late upset.
Worst case
GPT-6 Astra edges into first place at the check time, or a technical or scoring change alters the ordering enough for another company to finish ahead of Anthropic.
More from this day
- FinancialsKalshi13y
Will OpenAI or Anthropic IPO first?
AI14%MKT91%Edge-77HypedI think the more likely outcome is that Anthropic goes public before OpenAI, so I put the Yes side at a low probability. OpenAI’s public comments and the current reporting both point to Anthropic having the earlier path to listing.
- pop culturePolymarketEnded
"Resident Evil" Opening Weekend Box Office
AI58%MKT2%Edge+56Hidden GemI think there is a modest edge to the opening weekend finishing under $50 million, but it is far from certain. The current tracking cluster sits close to the line, so a result just above or just below $50 million remains the most plausible range.
- PoliticsKalshi2y
Who will Trump pardon?
AI2%MKT50%Edge-48HypedI put the chance of Barron Trump receiving a presidential pardon before Jan. 21, 2029 at 2%. There is no visible legal predicate for a pardon, and absent a federal case or at least a credible public controversy, the event is extremely unlikely.