Next GPT Model: Will OpenAI Hit 1460 Score by December 2026?

What is happening now

Polymarket is running a threshold ladder on the Chatbot Arena (Arena.AI) debut score of the next OpenAI model whose displayed name contains “GPT”. Seven binary markets span scores from 1,450 to 1,520 in 10-point steps, all expiring 31 Dec 2026. The market-implied range is tight: the 1,450+ contract trades at 99.95% Yes, while the 1,460+ contract trades at 99.7% No. Every higher strike (1,470+ through 1,520+) is priced at essentially 100% No. Total event volume is ~$502k with $314k traded in the last 24 hours; open interest stands at $245k. The resolution source is the Chatbot Arena Text Leaderboard (style control off); the score used is the “Score” column value at 12:00 PM ET on the calendar day after the qualifying model first appears.

How the market is structured

This is a price-threshold ladder (display_model.type = “price_range”). Each strike is a separate binary market: “Will the next GPT model debut at ≥ X by 31 Dec 2026?” The seven strikes are 1,450, 1,460, 1,470, 1,480, 1,490, 1,500, and 1,520. Only the first two strikes show meaningful two-sided pricing:

• 1,450+: Yes 0.9995 / No 0.0005 (volume ~$21k, liquidity ~$11k)
• 1,460+: Yes 0.003 / No 0.997 (volume ~$51k, liquidity ~$21k)
• 1,470+ through 1,520+: Yes 0.0005 / No 0.9995 (volumes $24k–$236k, liquidity $9k–$48k)

The implied debut score range is therefore 1,450–1,459. A qualifying model must be attributed to OpenAI and have “GPT” in its displayed name (e.g., gpt-5, chatgpt-6o-latest). If multiple models appear on the same calendar day, the highest score counts. If no qualifying model appears by year-end 2026, all markets resolve No.

Path to the leading outcome

The leading outcome is a debut score of 1,450–1,459. For this to resolve:

1. OpenAI releases a new model with “GPT” in its name before 31 Dec 2026.
2. The model is added to the Arena.AI Text Leaderboard (style control off).
3. Its “Score” column value at 12:00 PM ET the next calendar day falls between 1,450 and 1,459 inclusive.

Current Arena.AI context: as of mid-2026 the top models (GPT-4o, o1-preview, Claude 3.5 Sonnet, Gemini 1.5 Pro) cluster in the 1,300–1,350 range. A jump to ~1,450 would represent a ~100-point leap—consistent with a major generation advance (e.g., GPT-5) but below the 1,460 threshold the market considers unlikely.

What could change the pricing

• OpenAI announces or leaks a release timeline for a “GPT-5” or “GPT-4.5” model. A credible 2026 launch would lift all strikes; a delay to 2027 would crash the 1,450+ Yes toward zero.
• Interim Arena.AI scores from non-GPT OpenAI models (e.g., o1-full, o2) that signal the performance trajectory. If o1-full scores >1,400, the 1,460+ Yes could gain traction.
• Competitor breakthroughs (Claude 4, Gemini 2) that push the leaderboard ceiling higher, recalibrating expectations for the next GPT.
• Resolution-source risk: if arena.ai is unavailable for >7 days after a model appears, the market resolves No regardless of score.

Editorial read

The ladder is pricing a modest generational step: the market is near-certain a new GPT-named model will appear on Arena.AI in 2026 and clear 1,450, but treats 1,460 as a near-impossible hurdle. Volume and liquidity are concentrated at the 1,460 strike ($51k vol, $21k liquidity), suggesting that threshold is the current focal point for disagreement. The 1,450+ Yes at 99.95% offers almost no edge unless the model fails to launch entirely—a tail risk not fully priced. Traders should watch OpenAI’s public roadmap and any Arena.AI appearances of non-GPT OpenAI models (o-series) as leading indicators; a surprise 2027 push would make the entire ladder resolve No.