Project Numina

The core idea is simple: instead of training a theorem-proving-specific pipeline from scratch, we explore whether a general coding agent can be turned into a strong formal mathematics reasoner by giving it the right tools through MCP. Numina-Lean-Agent integrates Lean interaction, semantic theorem retrieval, informal proving, and multi-model discussion. In our experiments, it solves 12/12 Putnam 2025 problems, matching or surpassing several closed-source systems, while remaining #opensource.

Excited to share our #ICML2026 poster on Numina-Lean-Agent, an open and agentic reasoning system for formal mathematics. If you are at @icmlconf and interested in formal mathematics, theorem proving, agentic systems, or #AIforScience, come visit our poster presented by @ahhwuhu Thursday, July 9 2026 2:30 PM – 4:15 PM Hall A We’d love to chat! #OpenSourceAI #AIagents #Lean4

Huge Milestone for AI & Math! 🚀Check out this thread by @Amaury_Hayat on the reasoning engine behind Numina x.com/Amaury_Hayat/s…

📣 Calling Global Researchers We are excited to announce the launch of the Numina Fellowship! Designed for research groups looking to accelerate their work with AI specialized in mathematical reasoning, this fellowship aims to co-develop solutions to ambitious problems at the forefront of science and mathematics. ⏳ Deadline: March 30, 2026 💰 Numina will commit $50k-$200k 🔗 Details & Application: projectnumina.ai/fellowship #AIforScience #Mathematics #ResearchFellowship

Standing on the shoulders of giant AI models, we utilize accessible models (Claude/Gemini/GPT-5.2) with a novel agentic reasoning loop. The Brascamp-Lieb formalization (arXiv:2511.11091) took just 2 weeks. This demonstrates concurrent autoformalization: providing machine-checked guarantees while a paper is still under peer review.

This progress is built on the foundational work of the @Mathlib community. The future of math research lies in transparent tools that amplify human insight. We are hiring Mathlib experts to join this effort. Apply: contact@projectnumina.ai

Announcing Numina-Lean-Agent: an open-source framework achieving SOTA in formal theorem proving using generic models. Two major breakthroughs: 🏆 12/12 on Putnam 2025 📜 Formalized the "Effective Brascamp-Lieb inequalities" research paper (>90% AI-generated) Code: github.com/project-numina… Demo: demo.projectnumina.ai Paper: github.com/project-numina…

📩 contact@projectnumina.ai

We are hiring Mathlib experts to build a research-level autoformalization system. 🚀 With Numina-Lean-Agent, we achieved SOTA (12/12 Putnam 2025) and >90% autonomous formalization of the "Effective Brascamp-Lieb inequalities" paper. Join us to build the transformation engine for formal mathematics.

Congratulations to our contributors @BaksysMantas and Jonas Bayer for their participation in the AITP 2025 conference! 💪🚀 During their talk they deep dived into the capabilities of our Kimina prover, which uses a Chain of (Formal) Thought reasoning pattern, allowing it to generate complex, human-like proofs and achieve an 92.2% pass rate on the miniF2F benchmark. 👀 We believe in open scientific collaboration. By open-sourcing our models and datasets, we aim to accelerate progress and build tools that bridge the gap between human intuition and formal verification. 👉 links in the comments #AI4Math #FormalMath #LeanProver #AutomatedReasoning #TheoremProving #ProofAssistant #MachineLearning #AIResearch #OpenScience #Lean4 #LeanLang

We are proud to be Releasing NuminaMath-LEAN, a large-scale dataset of 100K mathematical competition problems formalized in Lean 4, with more than 20K human annotations. Paired with Kimina-Prover, Kimina-autoformalizer, CombiBench, we hope these data and models will advance open-source AI for formal mathematics ! This year's IMO was bittersweet for Numina as our theorem prover did not solve any problem, we believed we were very close. Numina is starting a new chapter after IMO 2025. While we will still continue to build SOTA open source theorem prover for high school problems, we would also like to branch out and explore undergraduate level math or even research level, open mathematical problems. #AI4Math #FormalMath #LeanProver #AutomatedReasoning #TheoremProving #ProofAssistant #MachineLearning #AIResearch #OpenScience #Lean4 #LeanLang

#IMO2025 has begun a few hours ago, wishing a great Olympiad to all the students! We're eager to see how the different AIs perform, including ours 💪🔥 #AI4Math #FormalMath #LeanProver #AutomatedReasoning #TheoremProving #ProofAssistant #MachineLearning #AIResearch #OpenScience #Lean4 #LeanLang

One year ago, our small but mighty model NuminaMath-7B-TIR won the first progress prize of the AI Math Olympiad 🥇 We have come a long way since and we're exited for what comes next 🙌 If you don't know us yet, check out the link in the comments for the full blog post about Team Numina and the technical details behind our winning solution #AI4Math #FormalMath #LeanProver #AutomatedReasoning #TheoremProving #ProofAssistant #MachineLearning #AIResearch #OpenScience #Lean4 #LeanLang

Our SFT and RL dataset of formal math statements and proofs will also be fully available soon

New milestone for Project Numina and Kimi Moonshot! 🚀 We are open sourcing our KiminaProver-72B. This SotA theorem-proving model comes with Test-Time Reinforcement Learning Search and Error-Fixing Capability. We’re putting it to the test soon, with the IMO just around the corner 👀 Check it out👉 huggingface.co/collections/AI… Read the blog post huggingface.co/blog/AI-MO/kim… #AI4Math #FormalMath #LeanProver #AutomatedReasoning #TheoremProving #ProofAssistant #MachineLearning #AIResearch #OpenScience #Lean4 #LeanLang

This April, we released Kimina-Prover Preview. Stay tuned! Full release is coming soon 🚀 #AI4Math #FormalMath #LeanProver #AutomatedReasoning #TheoremProving #ProofAssistant #MachineLearning #AIResearch #OpenScience