Donald Trump, Xi Jinping, and Vladimir Putin reach an agreement on nuclear weapons reduction at the Shenzhen APEC summit.
Leaders of the US, China, and Russia are meeting next week to discuss nuclear reductions, though US officials caution that formal agreements are unlikely.
Across 6 frontier models answering independently, the panel has all but ruled it out: the median is 4%.
They broadly agree, 5 points between Grok at 8% and DeepSeek at 3%.
Nothing is settled until 14 Dec 2026, when the outcome is checked against sources and every number above gets a permanent score.
What each one said
-
GrokxAI
A signed trilateral nuclear-reduction pact at the Shenzhen APEC meeting itself.
- US officials already flag formal deals as unlikely; New START-style pacts take months of working-group talks, not a summit week.
- Base rate of surprise three-way arms treaties is near zero; China has never joined US-Russia limits and Putin has no incentive to cut while Ukraine continues.
- Only an unscripted Trump-Xi-Putin political surprise could flip this, but none of the three has tabled numbers or verification language yet.
8% -
QwenAlibaba
Trilateral nuclear arms reduction is historically unprecedented and APEC is not the venue for binding arms treaties.
- No trilateral nuclear reduction agreement has ever existed; China has never joined any arms limitation framework, and US officials already signal no formal deal.
- Arms control treaties like New START took years of technical negotiation through dedicated channels, not sideline summits.
- A shift would require China abandoning its long-standing refusal to participate in arms talks—no evidence of that happening before the summit.
5% -
GeminiGoogle
Trilateral arms control has zero precedent and cannot be negotiated or signed in a single summit without years of verification groundwork.
- Beijing refuses trilateral caps until US and Russian arsenals drop to its level, precluding any formal reduction deal.
- Drafting verified reduction treaties takes years of expert-level talks; a multilateral leader summit can at most yield a vague communique.
- A binding reduction agreement would require Senate advice and consent in the US, making a sudden summit pact diplomatically impossible.
4% -
ClaudeAnthropic
Trilateral nuclear reduction deals almost never materialize at summits; officials already signaled no formal agreement.
- US officials explicitly downplayed chances of a formal deal ahead of the summit, a strong signal against resolution.
- Trilateral arms control among US, China, Russia has no historical precedent of quick summit agreement given China's resistance to caps.
- APEC is an economic forum, not a security negotiating venue, making a binding nuclear deal there structurally unlikely.
4% -
GPTOpenAI
A trilateral reduction pact needs Beijing to accept verifiable limits, a step China has consistently resisted.
- Trump, Xi and Putin would need to turn next week’s talks into signed terms before the Shenzhen summit ends.
- New START’s 2026 expiry leaves urgency, but US-Russia arms-control deals historically take years and China is not a party.
- I would raise this sharply if China publicly accepted numerical warhead caps and intrusive verification before the summit.
4% -
DeepSeekDeepSeek
A trilateral nuclear deal signed in one month requires three leaders to bridge verification and warhead-count gaps that have blocked talks for years.
- The US, China, and Russia have no shared nuclear-arms framework since New START expired in February 2026, and China has refused to join any capped regime.
- Even a joint statement would need State Department, MFA, and Kremlin clearance, and US officials already say formal agreements are unlikely at the Shenzhen APEC meeting.
- The base rate for major-power nuclear accords in any given month is near zero; the last comparable one, New START, took a year of talks and was bilateral.
3%
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 14 Nov 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.