Automated Theorem Proving on FormalML-Hard (Machine Learning Theory) 1.0 (test)
0.4Output Tokens (k)DreamProver
Evaluation Results
| Method | Links | |
|---|---|---|
| DreamProverSystem Category=Lemma Learning System, Base Model=Gemini 3.1 Pro2026.04 | 0.4 | |
| DreamProverSystem Category=Lemma Learning System, Base Model=Gemini 2.5 Pro2026.04 | 0.36 | |
| Gemini 2.5 ProSystem Category=Proprietary LLMs, Base Model=Gemini 2.5 Pro2026.04 | 0.29 | |
| Gemini 3.1 ProSystem Category=Proprietary LLMs, Base Model=Gemini 3.1 Pro2026.04 | 0.28 | |
| DreamProverSystem Category=Lemma Learning System, Base Model=GPT-5.3-Codex2026.04 | 0.23 | |
| GPT-5.3-CodexSystem Category=Proprietary LLMs, Base Model=GPT-5.3-Codex2026.04 | 0.1 |