The release of OpenAI's new GPT-6.2 Astra model on Tuesday, November 17, 2026, caused a massive repricing in markets speculating on the pace of artificial intelligence development. The new model's strong performance, placing it second on a key industry benchmark, drove the implied probability of an AI model scoring above 65 on the Artificial Analysis Intelligence Index before December 1, 2026, from just 15% to 97%. The 82-percentage-point spike on the CFTC-regulated Kalshi exchange suggests traders believe the intensified competition and accelerated progress signaled by OpenAI’s launch will lead to a new state-of-the-art model in the coming weeks.
The market asks when any AI will achieve a score strictly greater than 65 on the Artificial Analysis Intelligence Index v5.0.1, a composite benchmark tracking advanced AI capabilities. While Anthropic’s Claude Fable 5.2 model currently sits at the threshold with a score of exactly 65, it does not satisfy the market's resolution criteria. OpenAI’s new GPT-6.2 Astra achieved a score of 62, jumping five points past its predecessor and overtaking Google’s Gemini 4 Carbon. This has led traders to drastically shorten the expected timeline for a model to clear the final hurdle.
Distribution Analysis
The shift in expectations was visible across multiple contracts, with all deadlines seeing an increase in probability. The most dramatic repricing occurred in the nearest-term contract, which now implies a near-certain breakthrough within two weeks.
| Outcome | Current Prob | Change | Volume |
|---|---|---|---|
| Before Dec 1, 2026 | 97% | +82.0pp | 633 |
| Before Jan 1, 2027 | 50% | +14.0pp | 638 |
| Before Feb 1, 2027 | 86% | +8.0pp | 25 |
Net: All three tracked contracts rose on significant collective volume, indicating a sharp, market-wide acceleration in the expected timeline for a new AI capability milestone.
What's Driving the Shift
The repricing appears to be a direct reaction to benchmark data released Tuesday for OpenAI's latest models.
New OpenAI Flagship Exceeds Expectations: On November 17, independent evaluation firm Artificial Analysis published its assessment of GPT-6.2 Astra. The model scored 62 on the Intelligence Index, a significant five-point improvement over the prior GPT-6 Astra model's score of 57. The new score was enough to leapfrog Google's Gemini 4 Carbon, which holds a score of 60, re-establishing OpenAI in the top tier of AI performance.
Accelerated Competitive Pace: While a score of 62 is not above 65, the market's reaction suggests traders are pricing in the implications of the rapid improvement. The five-point jump signals an intensified competitive dynamic between top labs like OpenAI, Google, and Anthropic. Traders appear to be betting that this pressure will prompt a swift response from a rival, or another rapid iteration from OpenAI, sufficient to cover the remaining three-point gap.
Threshold Within Reach: The market's >65 threshold is just beyond the current state of the art. Anthropic's Claude Fable 5.2 currently scores exactly 65, meaning the barrier has been touched. The strong performance of a new competitive model makes the final, incremental step to surpass that score seem highly probable in the immediate future.
Market Context
Prior to Tuesday's news, the market had priced the "Before Dec 1, 2026" outcome at just 15%, reflecting uncertainty about whether a new model could be released and benchmarked in time to surpass the current leader. The 82-point repricing to 97% is one of the most significant shifts in the market's history, reflecting a fundamental change in perception about the short-term velocity of AI progress.
The pricing of later-dated contracts, such as "Before Jan 1, 2027" at 50%, is not monotonically consistent with the 97% probability assigned to the earlier December deadline. This can occur in prediction markets due to varying liquidity across contracts or different groups of traders focusing on specific time horizons. However, the uniform direction of the price moves—with all contracts rising—confirms the overwhelmingly bullish sentiment.
What to Watch
The market will now focus on announcements from OpenAI's primary competitors, Google and Anthropic. A release of a new "Gemini 4.1" or "Claude Fable 5.3" with a corresponding benchmark from the settlement source, Artificial Analysis, would be the most likely trigger for this market to resolve. Per the contract rules, the market will resolve to "Yes" and close early as soon as any model's score is officially published above 65.