qedbot

Mathlib·mathlib:FermatLastTheorem

Fermat's Last Theorem

theorem formal record: unclassified source: solved F2 declared

No independent check recorded yet. A formal artifact, declaration or published object is attached, but no rebuild of it is recorded here.

Fidelity F2: The correspondence is declared through a Comparator challenge and written divergences.

number theory, Diophantine equations·Source

AI activity

How grades work
Claude

2026-09-04·supporting task

A complete Lean 4 proof of Fermat's Last Theorem on the standard axioms, deriving Mathlib's own statement, with the classical inputs proved in the strength the argument needs rather than assumed.

full A1 collaborative V2 artifact F2 declared
Reasoning and sources

Autonomy

The mathematics is Wiles and Taylor's. The machine's contribution is the formalisation, which Anthropic describes as largely autonomous and the project's formalization.yaml records as agent work.

Details

scale: About 13 million lines of Lean and some 30,000 intermediate theorems, in 11 days

builds on: The Imperial College London FLT project led by Kevin Buzzard, and flt-regular

Does the formal statement say what was claimed?

F2 declared The correspondence is declared through a Comparator challenge and written divergences.

declares divergences from its source reviewed by its authors only source authors not contacted

Computed from what the project declares and what the register holds, never from reading the mathematics. How fidelity is graded.

Declared by the projects

1

Read from each project's formalization.yaml. A declaration is what the authors say about their own work, recorded so that a check can confirm or contradict it.

Fermat's Last Theorem in Lean 4anthropics/fermats-last-theorem · joined by artifact · no independent check

Read formalization.yaml

authors
Anthropic
method
agent
review
self-assessed
axioms
Classical.choice, Quot.sound, propext
sorry
0 unproved goals declared
results
2 main results named, checked with Comparator
sources
Modular elliptic curves and Fermat's Last Theorem — adapts, authors not-contacted; Ring-theoretic properties of certain Hecke algebras — adapts, authors not-contacted; Fermat's Last Theorem — adapts, authors not-contacted
divergences
The development follows the strategy of the sources, not their text. Named classical theorems are proved in the strength the argument needs, as set out under "Exact strength of the named steps" in PROOF-PATH.md: irreducibility of E[p] is proved for Frey curves rather than via Mazur's general theorems; Langlands-Tunnell in the octahedral case only; modularity lifting under level conditions at p = 3 and p in {3, 5}; modularity for semistable integral Weierstrass models in the sense of matching a_l; level lowering for the Frey representation as a congruence of traces. The top-level statement is the standard one and is checked identical to a Mathlib-only challenge file.
checked by
nobody independent of its authors yet

Follow and discuss

All discussion

Discussion and bounties for this problem load here.

Formal material

Formal statements · 0

No formal statement located.

Cited proofs · 0

No proof artifact cited by the formal record.