ComparIAThe duels: two AI tools head-to-head
Each ComparIA duel pits two AI tools against each other on an identical 8-prompt timed protocol — sourced research, news, summarizing, writing, calculation, code, instruction following, grammar — on free tiers, with manual fact-checking of every answer. First published duel: Perplexity vs ChatGPT (September 23, 2026), won by ChatGPT 4.8/5 vs 4.5/5.
All tested duels
Duels are listed from newest to oldest. Each card shows the exact test date — free tiers evolve fast, and the date is part of the result.
Perplexity vs ChatGPT
Tested September 23, 2026 — 8 prompts, free tiers. ChatGPT wins 4.8/5 vs 4.5/5 thanks to its smoothness and no-account access; Perplexity dominates sourced research, code and French grammar.
Read the duel →Next duel in preparation: ChatGPT vs Le Chat (Mistral). All duels follow our published methodology (quality 40%, ease of use 20%, value 20%, reliability 10%, features 10%).
Why duels instead of isolated scores?
An absolute score says little. A duel does: submitting the same prompt to both tools on the same day, timing the answers, checking the facts and counting respected instructions produces a reproducible comparison — and concrete differences that a general ranking hides. That is how we documented Perplexity's no-account blocking, or its surprise win on Python code.
Duels complement our category comparisons: the 6 best AI assistants, AI writing tools and AI image generators.