# google-deepmind/superhuman architecture

> Architecture diagram of the GitHub repository google-deepmind/superhuman, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2026-11-13). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/google-deepmind/superhuman
- Repository: https://github.com/google-deepmind/superhuman

## Overview

This repository is a collection of mathematical reasoning projects, benchmarks, and research artifacts rather than one application. The clearest documented workflow is Aletheia iteratively generating, verifying, and revising research-math solutions; the IMO benchmark datasets support evaluation of AI mathematical reasoning. The tree also contains Lean formalization and solution collections, but the available README excerpt does not describe their internal workflows, so no implementation-level links are inferred.

Main entry points: [`imobench`](https://github.com/google-deepmind/superhuman/tree/main/imobench) (IMO Bench datasets), [`aletheia`](https://github.com/google-deepmind/superhuman/tree/main/aletheia) (Aletheia agent), [`imobench/answerbench.csv`](https://github.com/google-deepmind/superhuman/blob/main/imobench/answerbench.csv) (AnswerBench), [`imobench/proofbench.csv`](https://github.com/google-deepmind/superhuman/blob/main/imobench/proofbench.csv) (ProofBench)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_benchmarks["Evaluation benchmarks"]
  node_imo_bench[("IMO Bench datasets")]
  node_answerbench["AnswerBench<br/>[answerbench.csv]"]
  node_proofbench["ProofBench<br/>[proofbench.csv]"]
  node_gradingbench["GradingBench<br/>[gradingbench.csv]"]
end

subgraph group_research["Research agent"]
  node_aletheia["Aletheia agent"]
  node_aletheia_outputs["Research solutions"]
end

subgraph group_formal["Formal mathematics"]
  node_futuremath["FutureMath formalization"]
  node_erdos_hajnal["Erdős–Hajnal proofs"]
  node_leap_solutions["LEAP solutions"]
end

node_researcher(("Math researcher"))

node_imo_bench -->|"includes"| node_answerbench
node_imo_bench -->|"includes"| node_proofbench
node_imo_bench -->|"includes"| node_gradingbench
node_imo_bench -->|"evaluates reasoning"| node_researcher
node_researcher -->|"uses"| node_aletheia
node_aletheia -->|"generates and revises"| node_aletheia_outputs
node_aletheia -->|"verifies solutions"| node_aletheia

click node_imo_bench "https://github.com/google-deepmind/superhuman/tree/main/imobench"
click node_answerbench "https://github.com/google-deepmind/superhuman/blob/main/imobench/answerbench.csv"
click node_proofbench "https://github.com/google-deepmind/superhuman/blob/main/imobench/proofbench.csv"
click node_gradingbench "https://github.com/google-deepmind/superhuman/blob/main/imobench/gradingbench.csv"
click node_aletheia "https://github.com/google-deepmind/superhuman/tree/main/aletheia"
click node_aletheia_outputs "https://github.com/google-deepmind/superhuman/tree/main/aletheia"
click node_futuremath "https://github.com/google-deepmind/superhuman/tree/main/FutureMath"
click node_erdos_hajnal "https://github.com/google-deepmind/superhuman/tree/main/FutureMath/Erdos-Hajnal/ErdosHajnal"
click node_leap_solutions "https://github.com/google-deepmind/superhuman/tree/main/leap/solutions"

classDef toneNeutral fill:#f8fafc,stroke:#334155,stroke-width:1.5px,color:#0f172a
classDef toneBlue fill:#dbeafe,stroke:#2563eb,stroke-width:1.5px,color:#172554
classDef toneAmber fill:#fef3c7,stroke:#d97706,stroke-width:1.5px,color:#78350f
classDef toneMint fill:#dcfce7,stroke:#16a34a,stroke-width:1.5px,color:#14532d
classDef toneRose fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
classDef toneIndigo fill:#e0e7ff,stroke:#4f46e5,stroke-width:1.5px,color:#312e81
classDef toneTeal fill:#ccfbf1,stroke:#0f766e,stroke-width:1.5px,color:#134e4a
class node_imo_bench,node_answerbench,node_proofbench,node_gradingbench toneBlue
class node_aletheia,node_aletheia_outputs toneAmber
class node_futuremath,node_erdos_hajnal,node_leap_solutions toneMint
class node_researcher toneIndigo
```

## Components

### Evaluation benchmarks

- [IMO Bench datasets](https://github.com/google-deepmind/superhuman/tree/main/imobench): `imobench`
- [AnswerBench](https://github.com/google-deepmind/superhuman/blob/main/imobench/answerbench.csv): `imobench/answerbench.csv`
- [ProofBench](https://github.com/google-deepmind/superhuman/blob/main/imobench/proofbench.csv): `imobench/proofbench.csv`
- [GradingBench](https://github.com/google-deepmind/superhuman/blob/main/imobench/gradingbench.csv): `imobench/gradingbench.csv`

### Research agent

- [Aletheia agent](https://github.com/google-deepmind/superhuman/tree/main/aletheia): `aletheia`
- [Research solutions](https://github.com/google-deepmind/superhuman/tree/main/aletheia): `aletheia`

### Formal mathematics

- [FutureMath formalization](https://github.com/google-deepmind/superhuman/tree/main/FutureMath): `FutureMath`
- [Erdős–Hajnal proofs](https://github.com/google-deepmind/superhuman/tree/main/FutureMath/Erdos-Hajnal/ErdosHajnal): `FutureMath/Erdos-Hajnal/ErdosHajnal`
- [LEAP solutions](https://github.com/google-deepmind/superhuman/tree/main/leap/solutions): `leap/solutions`

### Other components

- **Math researcher**

## Connections

- IMO Bench datasets → AnswerBench: includes (evidence: `README.md`)
- IMO Bench datasets → ProofBench: includes (evidence: `README.md`)
- IMO Bench datasets → GradingBench: includes (evidence: `README.md`)
- IMO Bench datasets → Math researcher: evaluates reasoning (evidence: `README.md`)
- Math researcher → Aletheia agent: uses (evidence: `README.md`)
- Aletheia agent → Research solutions: generates and revises (evidence: `README.md`)
- Aletheia agent → Aletheia agent: verifies solutions (evidence: `README.md`)