How ChatGPT-maker OpenAI’s ranking tumbled in Betting Markets after GPT-5 launch event, and Google’s jumped
Prediction markets have delivered a swift and brutal verdict on OpenAI’s GPT-5 launch, with Google now commanding a commanding 77% probability of having the best AI model by month’s end on Polymarket. The shift is nothing short of dramatic: Before Thursday’s demo (August 7), OpenAI was sitting pretty at around 80% odds while Google trailed at a distant 20%. By the time Sam Altman finished his presentation, those numbers had completely flipped. Turns out, traders weren’t buying what OpenAI was selling.The timing couldn’t be more striking. As GPT-5’s livestream was ongoing on Thursday, Polymarket odds flipped in real time, OpenAI started with around 80% probability and ended below 20%, while Google surged from second place at 20% to capture an impressive 77% market confidence. This live market reaction essentially provided an unfiltered, real-money assessment of OpenAI’s demonstration as it unfolded.The pre-launch hype had been considerable. Sam Altman had made bold predictions about GPT-5, asking audiences “How many of you still think you’re going to be smarter than GPT-5?” and stating “I don’t think I’m going to be smarter than GPT-5.” He even compared going back to GPT-4 as feeling “quite miserable,” likening it to “when the iPhone went from those giant-pixel old ones to the retina display.”The data reveals a sobering reality about AI progress: Even marginal improvements are harder to come by at the frontier. On Artificial Analysis’s Intelligence Index, GPT-5 with maximum reasoning scored just 69, only one point higher than Grok 4’s 68, two ahead of o3’s 67, and four above Gemini 2.5 Pro’s 65. When the gaps between leading models are measured in single digits, the race becomes less about breakthrough capabilities and more about incremental optimization.
Gemini’s structural advantages fuel market confidence
Google’s betting market dominance stems from both current performance and strategic positioning. Gemini 2.5 Pro continues to lead the LMSys Chatbot Arena leaderboard, the specific benchmark used for Polymarket resolution, with style control disabled. This technical detail gives Google a structural edge that sophisticated traders have clearly recognized.More importantly, Google released Gemini 2.5 Pro five months ago in March, making it the oldest flagship model among current leaders and positioning the company as overdue for a major update. The betting markets essentially wagered that Google’s development cycle and rumored Gemini 3.0 model could quickly respond to GPT-5’s launch with superior capabilities.Google’s broader AI ecosystem also provides competitive advantages. The company’s custom TPU infrastructure, massive data resources, and integration with Search and other services create natural moats that pure-play AI companies struggle to replicate. When OpenAI’s GPT-5 failed to deliver the revolutionary leap many expected, markets quickly pivoted to the next likely winner: Google.
The New AI hierarchy
Current Polymarket odds paint a fascinating hierarchy that goes beyond marketing claims. Google’s 77% probability towers over all competitors, while OpenAI has crashed to just 18%, a remarkable fall for the company that essentially created the modern AI race. xAI’s Grok holds a modest 4% despite Musk’s vocal promotion, tied with Anthropic’s Claude at the same level.The betting data becomes even more telling when examining the also-rans. DeepSeek, despite recent advances in reasoning models and cost efficiency, registers minimal confidence below 1%. Other Chinese AI companies fare similarly poorly, suggesting markets remain skeptical about their ability to compete at the frontier level by year-end.These percentages go beyond just current capabilities, they signal informed speculation about development cycles, resource allocation, and competitive positioning. Google’s overwhelming market confidence indicates traders believe the company has been strategically holding back superior models, waiting for the right moment to reclaim AI leadership.The resolution mechanism adds another analytical layer: these markets will be settled based on specific LMSys Chatbot Arena scores, where human preferences determine rankings. This means betting success depends not just on raw benchmark performance, but on creating models that feel genuinely better to use, a criterion where Google’s current Gemini 2.5 Pro already excels.What emerges is a maturing AI landscape where the dramatic capability jumps of earlier generations may be giving way to more incremental competition. When betting markets can flip 60 percentage points during a single product demonstration, it suggests that the moats between leading AI companies remain surprisingly narrow, and that public confidence in who’s “winning” can shift as rapidly as the market prices themselves. The real test will come in the next few weeks, as Google and other competitors respond to GPT-5 with their own model releases.
