PodcastIntelligence Snacks · Weekly conversations about AI, software and business, from the team behind Normie Mode.Listen →
Tests explained · Test of real work

Professional tasks in banking, consulting and law

Tasks written by experienced professionals that take a person about two hours, using documents, spreadsheets, email and slides.

Run by

Mercor

How to read it

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

Used for

7 tasks on this site.

How the models did

Share of tasks done correctly · Higher is better

Claude Fable 5.1 leads, followed by GPT-6 Astra and Gemini 3.8 Flash.

Claude Fable 5.1 · Anthropic69%
GPT-6 Astra · OpenAI65%
Gemini 3.8 Flash · Google DeepMind64%
Claude Sonnet 5 · Anthropic55%
Kimi K3 · Moonshot AI51%
DeepSeek V4 Pro · DeepSeek47%
Gemini 3.6 Flash · Google DeepMind47%
Gemini 3.1 Pro · Google DeepMind35%
Claude Haiku 4.5 · AnthropicNot tested yet
Claude Opus 5.5 · AnthropicNot tested yet
DeepSeek V4.1 Flash · DeepSeekNot tested yet
Gemini 3.5 Flash-Lite · Google DeepMindNot tested yet
GPT-5.6 Luna · OpenAINot tested yet
GPT-5.6 Sol · OpenAINot tested yet
GPT-6 Luna · OpenAINot tested yet
GPT-6 Sol · OpenAINot tested yet
Source: Mercor via Epoch AI · CC BY 4.0

Where we use it

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

TaskHow close
Best AI for businessTests this task
Best AI for accounting and financeTests a related skill
Best AI for financial analysisTests a related skill
Best AI for Excel and spreadsheetsTests a related skill
Best AI for data analysisTests a related skill
Best AI for presentations and PowerPointTests a related skill
Best AI for legal questionsTests a related skill
Intelligence Snacks newsletter

The big AI ideas each week, in your inbox