List your company

Theorem Proving on ProverBench

5.5Proof LengthClaude 4.6 Opus
3.02819.71436.453.086Apr 29, 2026

Evaluation results

MethodLinks
Claude 4.6 OpusType=Proprietary LLM2026.045.5
Gemini 3.1 ProType=Proprietary LLM2026.046.6
Gemini 2.5 ProType=Proprietary LLM2026.0412.7
GPT-5.3-CodexType=Proprietary LLM2026.0412.8
DeepSeek-Prover-V2-7BType=Open-source LLM2026.0419.6
Goedel-Prover-V2-8BType=Open-source LLM2026.0421.7
DreamProverBackbone=GPT-5.3-Codex, Type=Lemma Learning System2026.0434.1
Goedel-Prover-V2-32BType=Open-source LLM2026.0441.8
DreamProverBackbone=Gemini 2.5 Pro, Type=Lemma Learning System2026.0441.8
HilbertBackbone=Gemini 2.5 Pro, Type=Agentic System2026.0445.8
DreamProverBackbone=Gemini 3.1 Pro, Type=Lemma Learning System2026.0451.7
HilbertBackbone=Gemini 3.1 Pro, Type=Agentic System2026.0465.7
HilbertBackbone=GPT-5.3-Codex, Type=Agentic System2026.0467.3