Tests explained · Test of real work

Competition maths

Problems in the style of a top US high-school maths competition.

Run by

How to read it

The share of tasks the model got right. The thin line, where shown, is the likely range.

Used for

2 tasks on this site.

How the models did

Share of tasks done correctly · Higher is better

Claude Fable 5.1, GPT-5.6 Sol and GPT-6 Astra are level at the top.

Claude Fable 5.1 · Anthropic100%

GPT-5.6 Sol · OpenAI100%

GPT-6 Astra · OpenAI100%

Gemini 3.8 Flash · Google DeepMind99%

DeepSeek V4 Pro · DeepSeek99%

GPT-5.6 Luna · OpenAI98%

Kimi K3 · Moonshot AI97%

Gemini 3.1 Pro · Google DeepMind96%

Claude Sonnet 5 · Anthropic95%

Gemini 3.6 Flash · Google DeepMind94%

Gemini 3.5 Flash-Lite · Google DeepMind71%

Claude Haiku 4.5 · Anthropic36%

Claude Opus 5.5 · AnthropicNot tested yet

DeepSeek V4.1 Flash · DeepSeekNot tested yet

GPT-6 Luna · OpenAINot tested yet

GPT-6 Sol · OpenAINot tested yet

Source: Epoch AI via Epoch AI · CC BY 4.0

Where we use it

Each task page labels this test by how closely it matches the task.

Task	How close
Best AI for studying and homework	Tests a related skill
Best AI for maths and science	Tests this task