Field test
Would you pick the right model?
Every brief here is a real job with real limits. Choose the instrument you would actually reach for, and we will tell you whether it holds — and exactly why it does or does not.
14 scenarios · 22 candidates · No sign-up
Fig.01 — One brief, three plausible routes
/QUIZ
Method
Read · Pick · Learn
Three moves per
scenario.
The test is not a trivia round. Each brief has one answer that survives every constraint, one or two that are defensible, and at least one that looks cheap until you read the fine print.
Read the brief
A real working situation with hard constraints — budget, latency, volume, privacy, modality. No toy prompts.
Pick your instrument
Three or four real candidates, each with its true context window, list price, and tier. Weigh them and lock one in.
Learn why
We grade the pick, name the strongest answer, and walk the trade-offs option by option — including the ones you skipped.
Coverage
Counted from the live scenario set
- Scenarios
- 14
- Candidate models
- 22
- Topics covered
- 27
- Hard briefs
- 01
Topics in rotation
- accuracy
- agentic
- business
- coding
- cost
- creative
- education
- everyday
- extraction
- finance
- high-volume
- latency
- legal
- long-context
- marketing
- math
- multilingual
- on-prem
- open-weights
- real-time
- reliability
- research
- summarization
- support
- translation
- vision
- writing
Difficulty runs 02 easy, 11 medium and 01 hard. Scenario order is shuffled on every run, so a rematch is never the same paper twice.
Begin
No backend · Nothing stored
Ready when you are.
Fully keyboard-playable: arrow keys move between candidates, number keys jump straight to one, Enter locks your answer in and advances.