The Shanghai-held IMO featured 666 contestants from various countries, with only seven achieving full marks, underscoring the problems' difficulty and the strict under-20 age limit.

Two Chinese tech firms, Huawei and Xiaohongshu, announced that their AI models achieved perfect 100% scores on IMO problems, marking the first time AI systems have scored full marks under the official judging process.

Analysts note a rapid acceleration in AI math capabilities, suggesting the frontier may have moved beyond traditional IMO-level problems.

RedNote, based in Shanghai and known for Xiaohongshu, plans to open-source dots-note-3.0 soon, though a timeline has not been announced.

This milestone follows a trajectory of AI progress at the IMO, with 2024 seeing milestones like Google's silver-medal score and problem-solving on four of six tasks, and 2025 models reaching gold-level scores though not beating the top human contestants.

Reporting indicates the AI teams received the exam prompts only after human contestants completed the competition, with responses submitted within a strict timeframe.

Dots-note-3.0 is the lightest model in RedNote's dots3 family and is in beta, with larger variants like jazz and aria built for different compute resources.

Dots-note-3.0 is currently in beta and is the smallest member of the dots3 family, while larger variants are designed for varied use cases and optimized algorithms.

Experts anticipate a public release of dots-note-3.0 with potential applications beyond mathematics, signaling broader AI impact on fundamental sciences.

The development highlights ongoing progress in AI benchmarks for reasoning and math, with growing use of academic benchmarks to track AI progress across domains.

The achievement could influence education in Uzbekistan and beyond by using AI to explain and solve complex problems and may accelerate mathematical research.

Seven of the 666 contestants achieved a perfect score, illustrating the rarity of full marks in a field known for rigorous, proof-based solutions.