deepseek-ai/DeepSeek-Prover-V2

Health v2 · maintenance only

Flags: no_releases no_license

Adoption not part of the score

1297 stars · 107 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

DeepSeek-Prover-V2 is an open-source large language model for formal theorem proving in Lean 4, trained via reinforcement learning with subgoal decomposition seeded by DeepSeek-V3. The repository provides model weights, the ProverBench dataset, and quick-start instructions for inference.

Use cases

• prove theorems automatically in Lean 4
• generate formal mathematical proofs with an LLM
• benchmark LLMs on formal math reasoning
• train or fine-tune a theorem-proving model
• decompose complex math problems into subgoals

When to choose

• you need automated or assisted formal proof generation in Lean 4
• you want an open model specialized in formal mathematics
• you need a benchmark dataset for theorem proving research

When to avoid

• you need general-purpose chat or coding assistance
• you work in informal mathematics without a proof assistant
• you cannot host large GPU models locally

Facets

library · maturity active

machine-learning llm-training reinforcement-learning nlp artificial-intelligence large-language-models mathematics deep-learning python theorem-proving lean4 formal-mathematics open-weights dataset proverbench gpu linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="deepseek-ai/DeepSeek-Prover-V2")