Formal Theorem Proving on Synthetic 20
14.5Proof LengthGPT-5.3-Codex
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-5.3-CodexMethod Category=Proprietary LLMs2026.04 | 14.5 | |
| DreamProver (Gemini 2.5 Pro)Method Category=Lemma Learning System, Backbone Model=Gemini 2.5 Pro2026.04 | 15.4 | |
| Gemini 2.5 ProMethod Category=Proprietary LLMs2026.04 | 16.9 | |
| DreamProver (GPT-5.3-Codex)Method Category=Lemma Learning System, Backbone Model=GPT-5.3-Codex2026.04 | 17.3 | |
| Gemini 3.1 ProMethod Category=Proprietary LLMs2026.04 | 19.2 | |
| DreamProver (Gemini 3.1 Pro)Method Category=Lemma Learning System, Backbone Model=Gemini 3.1 Pro2026.04 | 21.1 |