Goedel-Prover-V2 scales open theorem proving with self-correction

The system uses scaffolded data synthesis, verifier-guided iterative repair and model averaging.

What happened?

Goedel-Prover-V2-32B reported 90.4% miniF2F in self-correction mode and 86 solved PutnamBench problems under its reported evaluation.

Why it matters

It strengthened the case for compact open provers that spend computation on correction rather than only increasing parameter count.

Significance rationale: Goedel-Prover-V2 showed that scaffolded synthetic data and Lean-guided self-correction could deliver strong open theorem-proving performance with much smaller models.

Historical context

Primary evidence

Evidence & status at a glance

Significance
H2 · Field milestone
Verification
Paper released
Evidence level
E3 · Inspectable artifact
AI role
AI primary, human verified
Formal assurance
Machine-checked

How should these labels be read? →

Contribution & verification

Mathematical novelty
Benchmark result
Interface
Formal, Symbolic
AI contribution
The model generates Lean proofs and iteratively repairs them using compiler feedback.
Human contribution
Researchers designed staged synthesis, self-correction and model-merging procedures.
Verification
Paper released · E3 · Inspectable artifact · Machine-checked
Last reviewed
2026-09-11

Technical details

Systems
goedel-prover-v2, lean
Organizations
princeton-university, nvidia
Problems
minif2f, putnambench
Methods
self-correction, synthetic-data, formal-proof

Sources & artifacts

Verification history

  • 2025-08-05 · Paper released · E2 · Research record

    The Chronicle records this event from the cited primary or official source; the status does not imply independent Chronicle verification.

  • 2026-09-11 · Paper released · E3 · Inspectable artifact

    This corpus refresh added a directly inspectable public artifact maintained by the authors or releasing organization. E3 records artifact-level inspectability and does not imply independent verification of the underlying scientific or mathematical claim.

Corrections

No corrections recorded.