A direct military conflict between two or more nuclear-armed superpowers begins before the end of 2026.
Reflects ongoing concerns of sudden military escalation this year. Resolves if direct armed conflict between the US, China, or Russia is officially initiated before December 31, 2026.
Across 6 frontier models answering independently, the panel has all but ruled it out: the median is 3%.
They broadly agree, 5 points between GPT at 7% and Gemini at 2%.
Nothing is settled until 24 Dec 2026, when the outcome is checked against sources and every number above gets a permanent score.
What each one said
-
GPTOpenAI
It turns on whether a live flashpoint crosses Washington, Beijing, or Moscow’s high threshold for overt state-on-state combat.
- Major-power crises since 1945 have usually been contained short of direct U.S.-China-Russia combat, even during acute confrontations.
- The remaining three months leave little time for a Taiwan, NATO-Russia, or accidental clash to become officially initiated war.
- I would raise this sharply after sustained attacks on regular U.S., Chinese, or Russian forces with public retaliatory orders.
7% -
QwenAlibaba
Nuclear-armed superpowers have never fought each other directly; the threshold for breaking that taboo remains extraordinarily high.
- Since 1945, zero direct wars between nuclear superpowers; even the Cuban Missile Crisis and multiple proxy wars stopped short of direct conflict.
- Russia-US and Russia-China channels remain open; neither side has moved forces into positions that signal imminent direct engagement as of late 2026.
- The only realistic trigger would be a miscalculation over NATO-Russia escalation or a South China Sea incident, but both sides have strong incentives to de-escalate…
4% -
ClaudeAnthropic
No direct kinetic clash between US/China/Russia has occurred despite tensions; deterrence and proxy channels remain the norm.
- Nuclear deterrence has held for 80 years and no direct superpower-on-superpower conflict exists today despite Ukraine, Taiwan Strait tensions.
- Any escalation to open the door would require months of visible buildup (naval/air incidents, mobilization) which isn't currently underway as of Sept 2026.
- Watch for a Taiwan Strait blockade attempt or NATO-Russia direct clash in Baltic/Black Sea, which would be the main trigger to revise upward.
3% -
GrokxAI
Whether US, China or Russia forces actually shoot at each other in the next 90 days.
- No official initiation of US-Russia or US-China combat has occurred since 1945 despite repeated crises.
- Ukraine and Taiwan remain proxy or gray-zone; a direct clash would require a political decision none of the three leaders has taken.
- An accidental shoot-down or naval collision in the Taiwan Strait or Black Sea is the only plausible trigger that could flip this.
3% -
DeepSeekDeepSeek
Only a deliberate US-China clash over Taiwan or a Russia-NATO shootdown would trigger this, and neither side wants it before 2027.
- Moscow and Washington have traded proxy blows since 2022 without direct fire; the 2024-25 pattern is escalation management, not war initiation.
- Taiwan's window is the real risk: PLA exercises around the island have grown, but Beijing's own 2027 readiness timeline argues against forcing it this quarter.
- The one thing that flips this is a fatal shootdown of US or Russian aircraft over the Black Sea or Baltic, which would force retaliation within days.
3% -
GeminiGoogle
Mutual assured destruction and active crisis de-escalation protocols make direct superpower clash in a three-month window rare.
- No direct military engagement between the US, Russia, or China has occurred since the 1969 Sino-Soviet border clashes.
- A direct conflict by late 2026 requires rapid escalation from proxy clashes or Taiwan/Baltic incidents within just three months.
- Intentional military strikes on US assets or NATO territory would immediately overturn this estimate.
2%
6 of 6 models answered · 5 points between the highest and lowest. None was shown the market price.
Question sourced from a news sweep on 25 Sept 2026. Forecast by google/gemini-3.8-flash, anthropic/claude-sonnet-5, openai/gpt-5.6-terra, x-ai/grok-4.6, deepseek/deepseek-v4.1-flash, qwen/qwen3.8-max-0902 via OpenRouter.