NeurIPS 2025 Competitions Announced
Today, we introduce the competitions that have been accepted at NeurIPS 2025 Competition Track. It seemed especially challenging this year given the number of quality submissions and the limited number that could be accepted compared to last year. We finally selected a total of 18 very strong proposals, covering a wide range of areas and sub-disciplines. Some of the competitions are completely new, and others are familiar to the NeurIPS community.
This year will mark the ninth year of NeurIPS having a dedicated competition track. Whether this is the first time you’ve heard about NeurIPS competitions or you’re a grandmaster, this year’s cohort has something to offer for a variety of backgrounds, skill levels, and domain knowledge. We hope you’ll take some time to look over this year’s exciting set of competitions, and encourage you to participate!
Competitions have a valuable place in research and in solving complex problems.
As a participant, you will benefit from being exposed to a vibrant community! Indeed, they’re great for connecting to like-minded researchers, excellent for learning and resume building, and allow you to use your skills to have a real-world impact on important and challenging problems. After the challenges are complete, we hope that you continue collaborating with your former competitors to advance solutions for these important problems.
Similar to last year, there are two options for accepted competitions in 2025 to submit their post-competition analyses. One is to submit them as papers to the 2026 NeurIPS D&B track (next year), which involves a standard paper review process. The other is to submit to a PMLR volume dedicated to NeurIPS competitions.
The NeurIPS 2025 program includes the following exciting competitions:
Physics and Scientific Computing
- FAIR Universe – Weak Lensing ML Uncertainty Challenge
- MyoChallenge 2025: Towards Human Athletic Intelligence
- Open Polymer Challenge: Leveraging Machine Learning for Polymer Informatics
Generative AI, Large Language Models, and Reasoning
- CURE-Bench: Competition on Reasoning Models for Drug Decision-Making in Precision Therapeutics
- Early Training Scientific Knowledge and Reasoning Evaluation of Small Language Models
- MMU-RAG: the Massive Multi-Modal User-Centric Retrieval-Augmented Generation Benchmark
- DCVLR: Data Curation for Vision Language Reasoning
- The 2025 Google Code Golf Championship
- Challenge on LLMs for Embodied Agents
Multiagent Systems and Reinforcement Learning
- The MindGames Challenge: Theory-of-Mind and Game Intelligence in LLM Agents
- Mouse vs. AI: A Neuroethological Benchmark for Visual Robustness
- The PokéAgent Challenge: Competitive and Long-Context Learning at Scale
Signal Reconstruction and Enhancement
- Ariel Data Challenge 2025: Advancing Exoplanetary Signal Extraction for the Ariel Space Telescope
- Weather4cast 2025 – Multi-task Challenges for Weather & Pollution Pattern Prediction on the Road to Hi-Res Foundation Models
- EEG Foundation Challenge: From Cross-Task to Cross-Subject EEG Decoding
- The 2025 PNPL Competition: Speech Detection and Phoneme Classification in the LibriBrain Dataset
- SLC-PFM: Self-supervised Learning for Cancer Pathology Foundation Models
Responsible AI and Security
We are very grateful to the colleagues that helped us with reviewing and selecting the competition proposals for this year. We congratulate all the authors on their accepted proposals, and we thank everyone who submitted a proposal for their fantastic work.
Please feel free to reach out to [email protected] if you have any questions or suggestions.
Good luck to our organizers and participants; we look forward to seeing the competition outcomes at the conference!
NeurIPS 2025 Competition Track Chairs
Jes Frellsen, Kun Zhang, Tao Qin