Five leading AI models—Kimi K26, ChatGPT 5.6 Sol, Claude Fable 5, Grok 4.3, and Gemini Pro—were asked to assess whether Bitcoin creator Satoshi Nakamoto was a sole individual or a group using Bayesian probability analysis. The models produced solo creator estimates ranging from 45% to 70%, a 25-point spread revealing more about AI reasoning patterns than about Satoshi's actual identity. Only ChatGPT 5.6 Sol reconciled its probabilities to a consistent 54% solo estimate, while Kimi K26 was the only model favoring a group explanation at 50% versus 45% for a sole creator. All five models cited consistent writing voice and coding style as their primary evidence base, yet four of five failed to properly reconcile their own Bayesian scenario percentages. The experiment highlights how AI systems structure uncertainty differently when confronting the seventeen-year-old mystery of Bitcoin's anonymous creator.
Five AI Models Assess Satoshi Nakamoto Identity Using Bayesian Analysis
Each AI model received an identical prompt asking it to build a simple Bayesian scenario tree, identify the three most likely scenarios for whether Satoshi Nakamoto was a sole individual or part of a group, assign probabilities to each scenario, and ensure the three scenarios plus an "other scenarios" category totaled 100%. The models were then asked to estimate the overall probability that Satoshi was a sole individual versus part of a group. A sole creator who has stayed silent for over a decade carries a different narrative and market weight than a small team that may still be active, coordinated, or holding keys under multiple control structures. Onchain patterns such as the Patoshi fingerprint have strengthened the case for a dominant early mining entity, yet they have not fully resolved whether that entity was one person or several working in concert.
Kimi K26 Favors Group Explanation at 50% Probability
Kimi K26 weighted sole authorship at 45% because the whitepaper and code exhibit a unified voice, consistent C++ style, and narrow expertise window typical of one polymath. However, the model assigned 35% probability to a small coordinated team scenario, citing the 2008-2010 round-the-clock forum activity and simultaneous cryptography, economics, and coding mastery. An institutional collective received 15% probability given the absence of leaks or bureaucratic artifacts, while "other" scenarios captured 5%. These priors yielded a 45% solo versus 50% group probability, making Kimi K26 the only model in the panel to favor a group explanation over a sole creator.
ChatGPT 5.6 Sol Produces Consistent 54% Solo Estimate
ChatGPT 5.6 Sol assigned 54% probability to a sole individual scenario, 30% to a small coordinated team, 10% to a lead creator with assistants, and 6% to other scenarios. The model was the only one in the panel to reconcile its scenario percentages back to a clean two-way split, showing its allocation work end-to-end. The 54% sole-individual estimate rested mainly on Satoshi's unusually consistent writing style, coherent technical philosophy, uniform coding approach, and tightly controlled public communications. The 46% group probability remained substantial because Bitcoin combined cryptography, distributed systems, economics, and software engineering at a level that could suggest several specialists working behind one identity.
Grok 4.3 Assigns 52% Probability to Sole Individual Scenario
Grok 4.3 produced a 52% posterior probability for a sole individual, starting from a 60% prior based on historical precedents of single cryptographers producing complex, cohesive systems. Primary evidence included stylometric analyses showing consistent voice and structure in the Bitcoin whitepaper, plus the Patoshi on-chain pattern evidencing unified early mining control by one entity. A small team scenario received 30% probability to accommodate breadth and slight forum-email stylistic shifts but was penalized by the clean multi-year absence of leaks or contradictions. The hybrid lead-plus-assistants case received 10% probability, while other scenarios received 8%.
Claude Fable 5 Assigns 50% Probability to Sole Author Scenario
Claude Fable 5 anchored on stylometric and behavioral evidence, assigning 50% probability to a sole author scenario. The model noted that Satoshi's writings show a consistent voice, spelling conventions, and idiosyncrasies including British spellings and double-spacing after periods, and the original codebase reflects a single programmer's style. However, the project's breadth in cryptography, economics, C++ engineering, and peer-to-peer networking exceeds typical individual expertise, and gaps in activity suggest possible consultation. The model reserved 25% probability for informal collaboration, 15% for a small coordinated team, and 10% for other scenarios. The model assumed public artifacts including emails, forum posts, and code are representative and not deliberately obfuscated.
Gemini Pro Assigns 70% Probability to Sole Individual Scenario
Gemini Pro produced the highest solo estimate in the panel at 70%, splitting the sole individual scenario into two sub-scenarios: an established cypherpunk (60%) and an isolated unknown (10%). The model's reasoning rested on three primary factors: the original codebase's stylistic consistency, the unified "voice" across emails, and timestamp data indicating a single person's sleep schedule. The core assumption was that a group would inevitably leak diverse coding habits or overlapping active hours. However, the model assigned 25% probability to a team scenario, acknowledging that a highly disciplined, small collective could theoretically mask their collaboration behind one persona. The 5% "other" category covered low-probability extremes.
AI Models Show 25-Point Spread in Satoshi Probability Estimates
The panel's numbers reveal more about model behavior than about Satoshi's identity. Five systems asked to run the same Bayesian exercise produced sole individual estimates ranging from 45% to 70%, a 25-point spread that undercuts any claim of AI consensus on the question. Only Kimi K26 broke from the pack entirely, favoring a group explanation over a solo Satoshi, while Grok, ChatGPT, and Claude Fable landed close together near the 50-50 mark. The gap exposes how loosely "Bayesian" gets applied: most models skipped the actual math, with four of five failing to reconcile their own scenario percentages back to a clean two-way sole versus group split. ChatGPT 5.6 Sol was the exception, showing its allocation work end-to-end. Gemini Pro's framing choice of splitting "sole individual" into two sub-scenarios likely explains its outlier 70% figure more than any unique evidence it cited. All five models pointed to the same two pillars—consistent writing voice and consistent coding style—as their evidence base, and all five drew an identical line between that evidence and pure speculation about specific identities or institutional backers.
FAQ
What probability did Kimi K26 assign to Satoshi being a sole individual?
Kimi K26 assigned 45% probability to a sole individual scenario and 50% probability to a group scenario, making it the only AI model in the panel to favor a group explanation over a sole creator.
Which AI model produced the most consistent Bayesian calculation?
ChatGPT 5.6 Sol was the only model to reconcile its scenario percentages back to a clean two-way split, showing its allocation work end-to-end with a consistent 54% solo estimate versus 46% group probability.
What evidence did all five AI models cite for their Satoshi assessments?
All five AI models cited consistent writing voice and consistent coding style as their primary evidence base, and all drew an identical line between that evidence and pure speculation about specific identities or institutional backers.
