At least one nuclear weapon is detonated in an act of war before 2030.
Fears of nuclear escalation remain acute amid regional conflicts. Resolution requires any offensive combat detonation of a nuclear warhead by a sovereign state or militant group.
Across 6 frontier models answering independently, the panel has all but ruled it out: the median is 6%.
They broadly agree, 15 points between GPT at 18% and Claude at 3%.
Nothing is settled until 9 Oct 2031, when the outcome is checked against sources and every number above gets a permanent score.
What each one said
-
GPTOpenAI
The key hinge is whether a major conventional war crosses a nuclear-use threshold before deterrence and crisis diplomacy contain it.
- Since 1945, nuclear weapons have not been used in war despite repeated severe crises, implying a low annual base rate.
- Russia, the US, China, India, Pakistan, North Korea, or a militant actor would need to authorize and execute use by October 2031.
- A direct NATO-Russia clash, India-Pakistan escalation, or Korean war would sharply raise this estimate; present conflicts remain below that threshold.
18% -
DeepSeekDeepSeek
The question turns on whether Putin, or another nuclear-armed leader facing existential battlefield defeat, breaks the 80-year taboo before 2030.
- Since 1945 only two wartime detonations occurred, both in 1945, implying a ~2.5% annual hazard; the 8-year window to 2030 yields roughly 18-20% baseline.
- Russia's war in Ukraine and Iran's nuclear program are live vectors, but Putin has not crossed the nuclear threshold despite battlefield setbacks, suggesting the…
- A single catalytic event, such as a Russian tactical nuclear use in Ukraine or an Iranian breakout test, would flip this from unlikely to near-certain.
15% -
GrokxAI
A nuclear-armed state must choose combat use despite 80 years of non-use and MAD.
- Russia, Pakistan, DPRK or Israel would have to authorize a warhead by end-2029 after conventional options fail.
- Post-1945 base rate is zero combat detonations in ~80 years; current flashpoints raise it only modestly.
- A tactical Russian strike in Ukraine or India-Pakistan escalation would most shift the odds.
8% -
GeminiGoogle
The taboo survives unless a nuclear power faces imminent regime decapitation or total military collapse.
- Zero wartime detonations have occurred across 81 years, setting a baseline annual hazard rate well below one percent.
- Russia, Israel, or North Korea would only strike if conventional defeat directly threatens regime survival, a high bar.
- Odds rise sharply if NATO directly enters Ukraine or the US and Iran engage in full-scale strikes on atomic facilities.
4% -
ClaudeAnthropic
No nuclear-armed state has detonated a weapon in combat since Nagasaki in 1945, and deterrence logic still holds.
- Base rate over 80 years of nuclear possession is effectively zero combat detonations outside 1945, even through Cold War, Kashmir, and current Ukraine/Mideast crises.
- Any use would require a state leader to accept near-certain severe retaliation or total pariah status, a threshold crises like Ukraine or Iran have not approached.
- Main risk path is India-Pakistan escalation or a collapsing North Korea/Russia scenario, but even acute crises historically resolve below the nuclear threshold.
3% -
QwenAlibaba
No state has used a nuclear weapon in war since 1945; the institutional and strategic barriers remain enormous.
- Base rate: zero combat detonations in 80 years despite multiple nuclear-armed rivalries and conventional wars (India-Pakistan 1999, 2019; Russia-Ukraine 2022-present).
- Russia's nuclear doctrine requires existential threat to the state; Ukraine war has stayed below that threshold for 4+ years, and NATO has deliberately avoided…
- Pakistan's doctrine is first-use against existential conventional threat, but India's Cold Start has never been executed at scale; a terrorist attack triggering full…
3%
6 of 6 models answered · 15 points between the highest and lowest. None was shown the market price.
Question sourced from a news sweep on 10 Oct 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.