Short Answer

Any AI model from any lab is likely to score above 65 on the Artificial Analysis Intelligence Index before December 1, 2026, with the market pricing this at 97.0%.

1. Market Behavior & Drivers

The market's probability surged 82.0 percentage points to 97.0% on November 17, 2026. The move directly followed OpenAI's announcement of its GPT-6.2 Astra and Sol models. The new GPT-6.2 Astra achieved a score of 62 on the Artificial Analysis Intelligence Index, a 5-point gain over the prior version and just three points below the market's resolution threshold of 65.
This sharp repricing reflects the proximity of multiple models to the target. At the time of OpenAI's release, another model, Claude Fable 5.2, was already scored at exactly 65. The market resolves if any model scores above 65 before the end of December 2026. With one model at the threshold and a key competitor demonstrating a significant jump to just below it, traders priced in a very high probability that a minor iterative improvement from any lab would trigger resolution.
  • Score above 65 likely before July 2027, Claude Fable 5.2 already at 65.
  • An imminent breakthrough is predicted; Google remains a leading dark horse.
  • Progress may plateau near 65 due to diminishing returns on advancements.

Who Wins and Why

Outcome Market Model Why
Before Dec 1, 2026 97.0% 91.3% Continued rapid advancements in AI capabilities likely enable reaching the score by December 1, 2026.
Before Jan 1, 2027 50.0% 53.0% Significant advancements are needed for AI to score above 65 before January 1, 2027.
Before Feb 1, 2027 86.0% 89.0% Current trends indicate AI systems are on track to exceed 65 before February 1, 2027.
Before Apr 1, 2027 90.0% 93.0% Accelerated progress in AI capabilities makes achieving the score probable before April 1, 2027.
Before Jul 1, 2027 91.0% 94.0% Ongoing research and development are expected to drive the score above 65 before July 1, 2027.

Current Context

The current top AI models approach, but do not strictly exceed, a 65 score on the Artificial Analysis Intelligence Index v5.0.1. Claude Fable 5.2 (Max, Default Fallback) holds the highest score at exactly 65 as of November 17-18, 2026 [1]. OpenAI released GPT-6.2 Astra and GPT-6.2 Sol on November 17, 2026; GPT-6.2 Astra (Max) scored 62, an increase of 5 points from GPT-6 Astra's 57, positioning it second behind Fable 5.2 and ahead of Gemini 4 Carbon (60) and Claude Opus 5.5 (59) [2]. GPT-6.2 Astra is priced at $8/$40 per million tokens, 20% below GPT-6 Astra [2]. The strictly greater than 65 threshold has not yet been crossed on the official AA leaderboard [1].
Prediction markets anticipate a greater than 65 score by early 2027. Kalshi prediction markets, opened October 16, 2026, with ~$20.7k volume, show odds of ~32% before January 1, 2027, ~48% before February 1, 2027, and ~64% before April 1, 2027 [3]. WhenTheModelDrop, updated November 17, 2026, provides a median forecast of February 8, 2027, with an 80% credible interval spanning December 9, 2026, to June 18, 2027 [3]. The AA Intelligence Index v5 launched on October 27, 2026, incorporating 9 evaluations, 60% of which use private datasets; Claude Opus 5.5 initially led at 58 at launch, with the top score since climbing to 65 [1].
Recent model releases indicate rapid progress in frontier AI capabilities. The top position on the AA Intelligence Index has changed hands several times in the past month. Gemini 4 Carbon briefly held the lead in late October 2026. Claude Fable 5.2 then reclaimed the top spot for Anthropic on November 9, 2026, before GPT-6.2 Astra's release pushed Carbon to third place on November 17, 2026 [4]. This suggests a fast cadence in frontier model development [4]. The 65-point threshold was met on prior index versions; a June 2026 report noted Claude Fable 5 scored 65 on the then-current index, and a Manifold market on whether the top AA score would reach 65+ by September 30, 2026 (on the older v4.x scale) resolved YES [5]. The v5 rescale reset scores lower [5].
Sources (5)
  1. 1Artificial Analysis Intelligence Index v5.0.1 | Artificial Analysisartificialanalysis.ai
  2. 2OpenAI has the #2 model in the Intelligence Index with the release of GPT-6.2 Astra and GPT-6.2 Sol | Artificial Analysisartificialanalysis.ai
  3. 3Artificial Analysis Index > 65 forecast — When The Model Dropswhenthemodeldrop.com
  4. 4GPT 6.2 Astra Places Second On Artificial Analysis Intelligence Index, Edges Out Gemini 4 Carbon By 2 Pointsofficechai.com
  5. 5Will the top Artificial Analysis Intelligence Index score reach 65 or higher on September 30, 2026? | Manifoldmanifold.markets

2. Price Chart

Historical Price (Probability)

Outcome probability
Date

3. Significant Price Movements

Notable price changes detected in the chart, along with research into what caused each movement.

Outcome: Before Dec 1, 2026

📈 November 17, 2026: 82.0pp spike

Price increased from 15.0% to 97.0%

What happened: The primary driver of the 82.0 percentage point price spike on November 17, 2026, was the announcement of OpenAI's GPT-6.2 Astra and Sol models [1]. This traditional news release revealed that GPT-6.2 Astra immediately scored 62 on the Artificial Analysis Intelligence Index, placing it second and demonstrating rapid progress towards the "strictly above 65" threshold required for the market to resolve YES by December 1, 2026 [1][2]. This significant development, occurring on the same day as the price movement, likely increased market participants' confidence that an AI model would soon exceed the 65-point mark. Social media was not identified as a factor in the provided information.

Outcome: Before Feb 1, 2027

📈 November 16, 2026: 11.0pp spike

Price increased from 67.0% to 78.0%

What happened: The primary driver of the price movement was the news surrounding OpenAI's GPT-6.2 Astra, which scored 62 on the Artificial Analysis Intelligence Index v5.0.1 [1]. This announcement, made on November 17, 2026, coincided with the 11.0 percentage point spike on November 16, 2026, as it placed a frontier AI model just four points away from the 66+ score needed by February 1, 2027 [3][4][5][1]. This represented a significant 5-point improvement from GPT-6 Astra, signaling rapid progress [1]. Social media's role as a primary driver is not explicitly detailed in the provided information; it was likely a contributing accelerant for the official announcement.
Sources (5)
  1. 1OpenAI has the #2 model in the Intelligence Index with the release of GPT-6.2 Astra and GPT-6.2 Sol | Artificial Analysisartificialanalysis.ai
  2. 2GPT 6.2 Astra Places Second On Artificial Analysis Intelligence Index, Edges Out Gemini 4 Carbon By 2 Pointsofficechai.com
  3. 3AA Intelligence Index above 65 before Feb 1, 2027? | Forecasting Agentforecast.agenticlearning.ai
  4. 4When will AI score above 65 on Artificial Analysis Intelligence Index? Odds & Predictionskalshi.com
  5. 5When will AI score above 65 on Artificial Analysis Intelligence Index? · PredictPalpredictpal.app

4. Market Data

Contract Snapshot

This market resolves based on when the Artificial Analysis Intelligence Index score for AI exceeds 65. For each sub-market, a 'YES' outcome wins if the score rises above 65 before its specified deadline (e.g., December 1, 2026, January 1, 2027, or February 1, 2027); otherwise, 'NO' wins. The final payout for the market is scheduled by July 1, 2027.

Available Contracts

Market options and current pricing

Outcome bucket Yes (price) No (price) Last trade probability
Before Dec 1, 2026 $0.19 $0.84 97%
Before Jan 1, 2027 $0.54 $0.51 50%
Before Feb 1, 2027 $0.96 $0.91 86%
Before Apr 1, 2027 $0.98 $0.47 90%
Before Jul 1, 2027 $0.98 $0.09 91%

Market Discussion

Traders largely anticipate AI scoring above 65 on the Artificial Analysis Intelligence Index relatively soon, with a strong 97% probability that this will occur before December 1, 2026. While opinions are split at 50% for it happening before January 1, 2027, there is an 86% chance it will happen before February 1, 2027. This indicates a strong consensus that the event is highly likely to occur by early 2027, with a significant expectation for it to happen even sooner, by the end of 2026.

5. Trust Index

Octagon Trust Index Kalshi 74 Good

Order book is critically thin.

Primary risk· Trade quality

How it adds up
Integrity80% of score84Good
Trade quality20% of score37High Risk

Includes the cost to trade: a $10,000 order can't be filled here because the order book is too thin.

Trust score74Good

Weighted blend with hard caps — a critically weak safety pillar, or a severe trading anomaly, caps the total regardless of the rest. Full methodology · About the Trust Index

Trust profile
Integrity4 screens run · 6 don't apply

6. Beyond OpenAI and Anthropic, which 'dark horse' labs like Google or Mistral could realistically release a model to surpass the 65-point threshold before July 2027?

Gemini Pro release likelihood88% likelihood for a near-term release [1][2]
Gemini 4 Carbon top-ranked odds (Nov 2026)73-80% odds to be top-ranked on November 30, 2026 [3][4][5][6][7][8]
Gemini first to 1550 on Text Arena57% to be the first model to hit 1550 on Text Arena before January 1, 2027 [4][6][9]
Google is the leading dark horse to surpass the 65-point threshold. Google is considered the most realistic dark horse laboratory likely to release a model surpassing the 65-point threshold on the Artificial Analysis Intelligence Index before July 2027 [4][6][9]. A key catalyst for this prediction is Google's next Gemini Pro model, which has an 88% likelihood for a near-term release [1][2].
Google demonstrated strong leadership in late 2026 prediction markets. In November 2026, Google's Gemini 4 Carbon (High) held the top position on the LM Arena leaderboard, with high odds (73-80%) of maintaining that rank through November 30, 2026 [3][4][5][6][7][8]. The company also consistently won 'Best AI this week' markets for several consecutive weeks in late 2026 [3][5][7][8][4][6][9]. Prediction market traders estimated Gemini had a 57% chance to be the first model to achieve a score of 1550 on Text Arena before January 1, 2027 [3][5][7][8][4][6][9]. Furthermore, Google was given a 76% probability to hold the highest Arena score at the close of 2026, positioning it as the de facto frontier leader outside of OpenAI and Anthropic [3][5][7][8][4][6][9].
While Google leads, other challengers exist, and the field narrows. xAI and DeepSeek are considered the most credible non-Google contenders in the race for a top-ranked AI model, though their probabilities as non-US labs were low [10]. The overall frontier AI competition through mid-2027 is widely anticipated to be dominated by new model releases from OpenAI, Anthropic, and Google [11][12][1][2].
Sources (12)
  1. 1AI Prediction Markets & Technology Odds | Kalshikalshi.com
  2. 2When will Anthropic release Opus 5.6? Odds & Predictions - Public.compublic.com
  3. 3Best AI this week Prediction Market Results - Public.compublic.com
  4. 4Top AI model this month Odds & Predictions - Public.compublic.com
  5. 5Best AI this week Odds & Predictions - Public.compublic.com
  6. 6Best AI this month Odds & Predictions - Public.compublic.com
  7. 7Best AI this week Prediction Market Results - Public.compublic.com
  8. 8Best AI this week Prediction Market Results - Public.compublic.com
  9. 9Which AI will be the first to hit 1550 on Text Arena? Odds & Predictions - Public.compublic.com
  10. 10Which companies will have a top-ranked AI model this year? Odds & Predictions - Public.compublic.com
  11. 11Which AI company will have the best coding model at the end of 2026? Odds & Predictions - Public.compublic.com
  12. 12When will OpenAI release Astra 6.3+? Odds & Predictions - Public.compublic.com

7. How has the rate of score improvement on the Artificial Analysis Intelligence Index v5 evolved since its launch in October 2026?

Current best model score approximationwithin roughly 1–2 points of 65 [1][2][3]
Expected time to cross 65approximately 2–6 weeks [1][2][3]
Median expectation for >65 scorelate December 2026 to early January 2027 [4][5][3][6][7][8]
Market odds predict an imminent breakthrough on the AI Index. While the precise rate of score improvement on the Artificial Analysis Intelligence Index v5 since its October 2026 launch is not explicitly detailed, market odds suggest the leading model is currently within approximately 1 to 2 points of reaching a score of 65 [1][2][3]. This threshold is anticipated to be surpassed within the next 2 to 6 weeks [1][2][3], reflecting growing anticipation for a significant breakthrough in AI capabilities [1][4]. As of November 18, 2026, official records do not yet indicate any AI model achieving an overall score strictly above 65 [2][3].
Recent market shifts indicate accelerating expectations for reaching 65. Specific market movements underscore these accelerating expectations. For instance, the Dec 1, 2026 contract experienced a notable +82 percentage point change in a 24-hour period, indicating a rapid re-pricing towards an imminent 65+ score [3][9]. Similarly, a snapshot from November 15, 2026, showed that lower odds for five related contracts subsequently rose over the following three days, with the Feb 1, 2027 contract gaining approximately 36 to 41 percentage points [1][4]. The median market expectation for an AI model to exceed a score of 65 is roughly late December 2026 to early January 2027, with Google's Gemini 4 Carbon (High) being identified as the probable candidate to achieve this milestone on the index [4][5][3][6][7][8].
Sources (9)
  1. 1When will AI score above 65 on Artificial Analysis Intelligence Index? | PredictPediapredictpedia.com
  2. 2AA Intelligence Index above 65 before Dec 1, 2026? | PredictPediapredictpedia.com
  3. 3When will AI score above 65 on Artificial Analysis Intelligence Index: Odds, Chart & News | 4Search Predictions4search.com
  4. 4When will AI score above 65 on Artificial… — Odds & Prices | DOXAdoxa.pro
  5. 5When will AI score above 65 on Artificial Analysis… Odds | Kalshipredictmarketcap.com
  6. 6Best AI this week Prediction Market Results - Public.compublic.com
  7. 7Best AI this week Prediction Market Results - Public.compublic.com
  8. 8Best AI this month Odds & Predictions - Public.compublic.com
  9. 9AA Intelligence Index above 65 before Dec 1, 2026?… | Frenzy Capitalfrenzycap.com

8. How do the sub-benchmark scores of Claude Fable 5.2 and GPT-6.2 Astra compare across the nine components of the AA Intelligence Index v5?

AA Index >65 Market Odds97% 'Before Dec 1, 2026' [1][2]
Fable 5.2 LM Arena Score (Above 1530)96% [3][4]
Astra 6.1+ LM Arena Score (Above 1460)52% [5][4]
AA Intelligence Index sub-benchmark data for frontier models is currently unavailable. Research indicates that information regarding a nine-component breakdown of the AA Intelligence Index v5, specifically for Claude Fable 5.2 and GPT-6.2 Astra, is not publicly accessible [6][2][1]. Furthermore, the existence of nine named components within this index could not be verified through available sources [6][2][1]. The only accessible index data pertains to overall-score thresholds, rather than specific per-component scores [6][2][1].
Prediction markets anticipate frontier models exceeding 65 on the AA Index. The market odds of 97% for 'Before Dec 1, 2026' on the >65 AA Index market suggest that traders expect one of the new frontier models, likely Fable 5.2 or Astra 6.1+, to achieve an official overall score above 65 within two weeks [1][2]. However, the specific model anticipated to reach this score and its component-level results remain unconfirmed by any available source [1][2].
Fable 5.2 generally outperforms Astra 6.1+ in early arena scores. Anthropic's Fable 5.2 was publicly released by mid-November 2026, with its debut LM Arena Text Score market indicating a 96% probability for 'Above 1530' and an 89% probability for 'Above 1540' [3][4]. OpenAI's Astra 6.1+ release market resolved 'Yes' for 'Before Dec 31, 2026' at 99% [5][4]. Its debut arena score market priced 'Above 1460' at 52% and 'Above 1480' at 41%, suggesting Astra's expected arena performance may trail that of Fable 5.2. Despite these general performance indicators, no AA Index sub-scores were found for either model [5][4].
Sources (6)
  1. 1When will AI score above 65 on Artificial… — Odds & Prices | DOXAdoxa.pro
  2. 2When will AI score above 65 on Artificial Analysis Intelligence Index? Odds & Predictions - Public.compublic.com
  3. 3When will Anthropic release Fable 5.2? Prediction Market Results - Public.compublic.com
  4. 4Fable 5.2+ debut arena score Odds & Predictions - Public.compublic.com
  5. 5When will OpenAI release Astra 6.1+? Prediction Market Results - Public.compublic.com
  6. 6When will AI score above 65 on Artificial Analysis Intelligence Index? | PredictPediapredictpedia.com

9. What are the key differences in evaluations and scoring methodology between the AA Intelligence Index v5 and its prior v4.x versions?

Number of Evaluations (v5.0.1)Exactly nine [1][2]
Evaluation Categories (v5.0.1)Four, each 25% weighted [1][2]
Key V5 Scoring EmphasisAgentic, tool-using evaluation [1][2]
The provided research for the AA Intelligence Index v5.0.1 does not detail prior v4.x versions. This absence of information about earlier iterations prevents a determination of key differences between v5 and its predecessors. The AA Intelligence Index v5.0.1 incorporates exactly nine evaluations, which are organized into four distinct categories, each weighted at 25%: Knowledge Work, Coding, Science, and General [1][2].
Nine specific evaluations comprise the AA Intelligence Index v5.0.1 scoring methodology. These include AA-Briefcase v1.1 (15%), AutomationBench-AA (5%), GDP.pdf (5%), FrontierCode-AA 2.0 (15%), Terminal-Bench 4.0 (10%), Terminal-Bench-Science 0.1.0 (15%), CritPt v2 Hard Subset (10%), AA-Omniscience v1.1 (15%), and HLE-Diamond (10%) [1][2].
Agentic, tool-using evaluations form a core v5 scoring emphasis across the index. This approach is exemplified by components such as AA-Briefcase v1.1, which employs pairwise comparisons on rubric-graded task success, analytical quality, and presentation quality, fitted with a Crowd-BT model. Coding and science evaluations like FrontierCode-AA 2.0 and Terminal-Bench 4.0 also adhere to this terminal or agentic methodology [1][2]. Standardized testing parameters for v5 include setting temperature at 0 for non-reasoning models and 0.6 for reasoning models, limiting non-reasoning models to a maximum of 16,384 output tokens, and utilizing zero-shot instruction prompting [1].
Sources (2)
  1. 1Intelligence Benchmarking | Artificial Analysisartificialanalysis.ai
  2. 2智能基准测试 | Artificial Analysisartificialanalysis.ai

10. What technical arguments or evidence suggest that AI model progress on the AA Index v5 might plateau before strictly exceeding a score of 65?

AA Index v5 Plateau Range64-65 range [1][2][3]
Text Arena 1550 Probability~56-59% [2][4][5]
Text Arena 1600 Probability~6% [2][4][5]
AI model progress on the AA Index v5 may plateau near 65. Technical arguments suggest that advancements could stall in the 64-65 range due to the discrete nature of progress between model releases [1][2][3]. This trend aligns with evidence from related capability markets, which indicates that frontier gains are slowing at the top end [2][4][5]. Market sentiment, reflected in non-monotonic odds, suggests traders currently perceive the best AA Index v5 score as just under 65, implying that the next qualifying release capable of crossing the 65 threshold might be months away if an imminent release does not succeed [1][2][3].
Structural and technical factors significantly contribute to this potential plateau. A critical factor is the market's strict threshold: a score of 64.9 fails to qualify, and a precise 65 does not resolve the market, as resolution relies on a single official published figure [6][1][7][6][1][8]. Furthermore, the AA Index disproportionately weights hard, frontier tasks where top models already cluster, meaning incremental gains on easier tasks result in diminishing overall score movement [6][1][8]. Finally, compute and power limitations, such as the requirements for multi-gigawatt datacenter rollouts and escalating memory inflation, are identified as potential delays for the next-generation training runs necessary to achieve a significant step-change in performance [6][1][8].
Sources (8)
  1. 1When will AI score above 65 on Artificial… — Odds & Prices | DOXAdoxa.pro
  2. 2When will AI score above 65 on Artificial Analysis Intelligence Index? · Before Jul 1, 2027 91% — Kalshi oddspdata.world
  3. 3AA Intelligence Index above 65 before Jan 1, 2027? | PredictPediapredictpedia.com
  4. 4AI Prediction Markets & Technology Odds | Kalshikalshi.com
  5. 5Anthropic market share this week Prediction Market Results - Public.compublic.com
  6. 6When will AI score above 65 on Artificial Analysis Intelligence Index? | PredictPediapredictpedia.com
  7. 7AA Intelligence Index above 65 before Dec 1, 2026? | PredictPediapredictpedia.com
  8. 8Stock Contextstockcontext.info

11. What Could Change the Odds

Key Catalysts

As of November 17, 2026, Anthropic's Claude Fable 5.2 (Max, Default Fallback) holds the lead on the AA Intelligence Index v5.0.1 at exactly 65, with GPT-6.2 Astra (Max) at 62 and Gemini 4 Carbon (High) at 60 [1][2][3][4]. The market's resolution condition requires a score strictly greater than 65; equality does not qualify [5]. The bullish outlook cites that only a +1 point increase is needed for resolution. Frontier labs consistently release flagship models every 1-3 months, as seen with Anthropic's Fable 5.2 (November 9, 2026) and OpenAI's GPT-6.2 Astra/Sol (November 17, 2026) [4][6][7]. Historically, on a stable index version, the leading model has risen 3-7 points per quarter [4].
The primary bearish risk stems from AA re-versioning. The October 2026 shift to Index v5.0 (v5.0.1 by November 7) re-based scores lower, with prior v4.x tops of ~66 before the rescale [4][8]. A potential v5.1 rescale could again push the top model's score below 65. Additionally, confirmed frontier release dates for December 2026 and January 2027 are not available [4]. Kalshi's 'When will AI score above 65 on AA Intelligence Index' market prices the probability before December 1, 2026 at ~97% [5][9]. The market resolves Yes if any model's official publicly displayed overall score on artificialanalysis.ai/leaderboards/providers is strictly greater than 65 [5].

Key Dates & Catalysts

  • Strike Date: July 01, 2027
  • Expiration: December 01, 2026
  • Closes: July 01, 2027
Sources (9)
  1. 1Artificial Analysis Intelligence Index v5.0.1 | Artificial Analysisartificialanalysis.ai
  2. 2Comparison of AI Models across Intelligence, Performance, and Price | Artificial Analysisartificialanalysis.ai
  3. 3AA Intelligence Index leaderboard: Claude Fable 5.2 leads 44 modelsmodelcap.ai
  4. 4AA Intelligence Index above 65 before Feb 1, 2027? | Forecasting Agentforecast.agenticlearning.ai
  5. 5When will AI score above 65 on Artificial Analysis Intelligence Index? Odds & Predictions - Public.compublic.com
  6. 6OpenAI has the #2 model in the Intelligence Index with the release of GPT-6.2 Astra and GPT-6.2 Sol | Artificial Analysisartificialanalysis.ai
  7. 7GPT 6.2 Astra Places Second On Artificial Analysis Intelligence Index, Edges Out Gemini 4 Carbon By 2 Pointsofficechai.com
  8. 8Will the top Artificial Analysis Intelligence Index score reach 65 or higher on September 30, 2026? | Manifoldmanifold.markets
  9. 9AI Prediction Markets & Technology Odds | Kalshikalshi.com

13. Related News