Stop guessingMatch tasks to the right AI

Describe what you need to finish. We combine public evaluations, product capabilities and your preferences to recommend three products and generate an adapted prompt.

No sign-inChina access notesPublic sources included

What do you want AI to help you complete?Plain language is fine

More detail leads to a better match0/800

Or choose a task type

What matters most?

This prototype classifies tasks locally in your browser and does not upload the description. The production version calls DeepSeek server-side without exposing the key in the page.

YOUR SHORTLIST

These three are worth trying first

Actionable recommendations for this task and preference.

My understanding

TRANSPARENT BY DESIGN

Where do recommendations come from?

The first release does not invent its own benchmark or present vendor claims as fact. We map multiple public sources into understandable task-fit levels while preserving dates and limitations.

Current data update · 2026-08-09Next planned update · 2026-09-01

Understand the task

DeepSeek structures Chinese and English task descriptions, with local browser rules as a fallback.

Match public evidence

We combine independent leaderboards, open evaluations, official capability notes and regional availability without mixing incomparable raw scores.

Re-rank by preference

Quality, balance, China access and speed use different weights; every card explains both strengths and limits.

Artificial AnalysisIntelligence · Speed · Price ArenaHuman preference LiveBenchObjective evaluation OpenCompassChinese benchmarks CompassArenaChinese preference OpenRouterModel metadata