In an extraordinary demonstration of automated reasoning, historical cryptanalysis, and linguistic deduction, evaluation benchmark lab Vals AI has revealed that Anthropic’s Claude Fable 5.1 has successfully solved the 'Cyphral Distich'—a cryptographic puzzle that had remained unsolved since 1653.
At the conclusion of Sir Thomas Urquhart’s 1653 treatise Logopandecteision, the eccentric Scottish polymath and Royalist included a cryptogram consisting of two lines of 32 numbers each. For over three centuries, the cipher resisted all attempts at decryption. It was formally posed as an open problem in scholarly journals including Notes and Queries, but human cryptanalysts were stymied by the brevity of the text and Urquhart’s famously baroque prose style.
Vals AI gave Claude Fable 5.1 the raw 64 numbers as an open-ended objective, without pre-programmed hints or external scaffolding. Rather than simply brute-forcing substitution combinations, Fable 5.1 approached the cipher with structured intellectual curiosity.
First, the model identified that the numerical values clustered around specific intervals that corresponded to the letters of the 24-letter English-Latin alphabet of the mid-17th century. Second, it recognized that Urquhart was an ardent Royalist who fought at the Battle of Worcester, and deduced that the numbers likely indexed a shared keyword derived from loyalist literature of the era.
Within hours, Fable 5.1 wrote small Python verification scripts to test candidate homophonic substitution tables against 17th-century Scottish orthography. By hour 22, the model converged on the valid key, outputting the coherent, rhyming English couplet hidden beneath the digits.
The result highlights a profound inflection point in AI capabilities: the emergence of machine reasoning systems capable of synthesizing domain knowledge across centuries of history, literature, and mathematics to crack problems that eluded human scholars for generations.