Skip to content
forward pass

Gold at the Mathematical Olympiad

Models from Google DeepMind and OpenAI solve five of six problems at the International Mathematical Olympiad in natural language, under contest conditions, matching the top human students.

category
culture
significance
4 of 5
people
Demis Hassabis, Alexander Wei
organisations
Google DeepMind, OpenAI

what had to happen · 58 events back to 1943

Every event this one built on, transitively, in order. The organism has the same path lit. Direct influences are marked.

00 · One neuron · 1

  1. 1943A logical calculus of nervous activity

I · Foundations · 11

  1. 1948A mathematical theory of communication
  2. 1949Cells that fire together wire together
  3. 1950Programming a computer for playing chess
  4. 1950Computing machinery and intelligence
  5. 1958The perceptron learns
  6. 1959Samuel's checkers program coins 'machine learning'
  7. 1960ADALINE and the least-mean-squares rule
  8. 1965Moore's law
  9. 1966ELIZA
  10. 1969Perceptrons
  11. 1970Reverse-mode automatic differentiation

W1 · The first winter · 2

  1. 1974Werbos applies backpropagation to neural networks
  2. 1980The Neocognitron

II · Connection · 3

  1. 1982The Hopfield network
  2. 1985The Boltzmann machine
  3. 1986Backpropagation

W2 · The second winter · 6

  1. 1988Temporal-difference learning
  2. 1989Q-learning
  3. 1989LeNet reads handwritten postcodes
  4. 1990Finding structure in time
  5. 1991The vanishing gradient problem
  6. 1992TD-Gammon reaches world-class backgammon

III · Statistics and data · 9

  1. 1997Long short-term memory
  2. 1998MNIST and LeNet-5
  3. 1999The first GPU
  4. 2003A neural probabilistic language model
  5. 2006Deep belief networks and the word 'deep'
  6. 2007CUDA
  7. 2009Deep learning moves to GPUs
  8. 2009ImageNet
  9. 2010Rectified linear units

IV · Deep learning · 9

  1. 2012Google Brain's network discovers cats
  2. 2012Dropout
  3. 2012AlexNet wins ImageNet
  4. 2013Deep Q-networks play Atari
  5. 2014Attention
  6. 2014Sequence to sequence learning
  7. 2015Batch normalisation
  8. 2015Residual networks
  9. 2016AlphaGo beats Lee Sedol

V · Transformers · 8

  1. 2017Attention is all you need
  2. 2017Deep reinforcement learning from human preferences
  3. 2017AlphaGo Zero learns from nothing
  4. 2018GPT: generative pre-training
  5. 2019GPT-2 and the model too dangerous to release
  6. 2020Scaling laws for neural language models
  7. 2020GPT-3
  8. 2020Learning to summarise from human feedback

VI · Everyone · 8

  1. 2022InstructGPT
  2. 2022Chain-of-thought promptingdirect
  3. 2022Chinchilla: the models were undertrained
  4. 2022ChatGPT
  5. 2023LLaMA leaks and open weights take off
  6. 2023GPT-4
  7. 2024o1 and reasoning modelsdirect
  8. 2024DeepSeek-V3 trained for $5.6 million

VII · Agents · 1

  1. 2025DeepSeek-R1direct

The International Mathematical Olympiad is the hardest examination that teenagers sit, six proof problems over two days, and in July 2025 it was held on Australia's Sunshine Coast. Google DeepMind entered an experimental version of Gemini with what it called Deep Think, working in ordinary English rather than a formal proof language, under the same four-and-a-half-hour limits as the students. It solved five of the six problems for 35 points, a gold medal, and the organisers graded and confirmed the result on 21 July. OpenAI had announced two days earlier that an unreleased model of its own had scored the same, graded by former medallists, and was criticised for pre-empting the students' ceremony.

A year earlier DeepMind's AlphaProof had won silver in the formal language Lean, with days per problem. The 2025 result was in prose, in time, and from general-purpose reasoning models rather than a system built for mathematics.

It is on this timeline as the moment a milestone that forecasters in 2021 had put at 2030 or later was passed, and as the clearest evidence that the reasoning methods of o1 and R1 generalised. Whether a machine that proves olympiad theorems can do mathematics remained, in the profession, an argument.

what it led to · 1 events downstream, through 2025

Built on it directly:

  1. 2025Gemini 3VII

sources · 2

Trace the lineage of this event on the timeline →Back to the ledger