Technology / 2 MIN READ
Traders now price just 1 percent odds that OpenAI's next model will debut above 1470, down sharply after Friday's Sol announcement.
Market data
Current live odds
Will the next model released by OpenAI debut at a score of at least 1470?
Technology / 1 MIN READ
The company released GPT-5.5 in April and has not confirmed a successor, but internal communications suggest a rapid iteration cycle under pressure from rivals.
Technology / 1 MIN READ
The company's most recent model, GPT-5.5, launched in April and its release notes show no sign of the next iteration.
Prediction markets now assign a 1 percent probability that OpenAI's next model release will debut at a benchmark score of 1470 or higher, following an 83-point decline in implied odds on Friday. The Polymarket contract dropped to 0.75 percent after OpenAI announced a limited preview of GPT-5.6 Sol, its flagship model in the new series, which the company says achieves 92.2 percent on BrowseComp and 62.6 percent on OSWorld 2.0. The market has traded $10,758 in the past 24 hours and remains open through December 31.
OpenAI unveiled three GPT-5.6 variants on July 11—Sol, Terra, and Luna—with Sol initially available only to select partners and organizations. The company also highlighted Sol's 73.5 percent score on ExploitBench, up from GPT-5.5's 47.9 percent at a comparable output-token budget. Despite the performance gains on those benchmarks, traders appear skeptical that the next numbered release will clear the 1470 threshold, with the Polymarket market page showing the 1450-plus outcome at 54 percent and 1470-plus trailing at 34 percent.
A related Polymarket contract on the next GPT model prices 1450-plus at 95 percent and 1460-plus at 85 percent, suggesting traders expect the debut score to cluster in the 1450 to 1460 range rather than reach 1470. OpenAI's model release notes show the company continues to ship new numbered models, including o3-pro, GPT-4.1 mini, and o4-mini, but the specific benchmark used by Polymarket to resolve the question remains undefined in public documentation. The sharp drop in odds reflects trader uncertainty over whether OpenAI will unveil an unusually strong model before year-end or whether the next release will land below the 1470 mark.
The market's collapse underscores the difficulty of forecasting benchmark scores months in advance, particularly when OpenAI's release cadence and evaluation criteria remain opaque. With five and a half months until the contract closes, traders are pricing in a low likelihood that the next model will exceed the 1470 threshold, even as OpenAI touts state-of-the-art results on multiple benchmarks for GPT-5.6 Sol.