We beat DeepL at its strongest settings. 99 to 3.
118 blind rounds across 11 country variants of Spanish, Portuguese, French, Arabic and Chinese, refereed by Google's Gemini 4 Argon. Every sentence, translation and verdict is published.
November 1, 2026AIRSkip Inc.Get the data
99
won by AIRTranslate
16
ties
3
won by DeepL
84% of 118 blind roundsJudged by Gemini 4 ArgonEvery round read in both orders
DeepL is the translator people upgrade to when Google Translate is not good enough. That is exactly why we picked it.
If you claim to translate the way locals actually speak, you test against the toughest opponent available, at its strongest settings, with a referee you do not control.
01The result
99 rounds to AIRTranslate. 3 to DeepL.
Each square below is one round: one sentence, asked for one country, answered by both systems and judged blind by Gemini 4 Argon. AIRTranslate was preferred in 99 of 118. DeepL was preferred 3 times. The other 16 were ties.
Tap any square to read that round
- AIRTranslate preferred
- Tie
- DeepL preferred
Spanish (Mexico)
12/12
Spanish (Argentina)
12/12
Spanish (Colombia)
11/12
Spanish (Spain)
8/12
Portuguese (Brazil)
5/10
Portuguese (Portugal)
8/10
French (France)
6/10
French (Quebec)
10/10
Arabic (Egypt)
10/10
Arabic (Saudi Arabia)
10/10
Chinese (Taiwan)
7/10
Spanish (Argentina) · sentence 10 of 12
“I'm broke until payday.”
DeepL
Estoy sin un peso hasta el día de pago.
AIRTranslateJudge's pick
Estoy sin un mango hasta que cobre.
JudgeGemini 4 Argon
"Sin un mango" and "hasta que cobre" are the exact idiomatic expressions an Argentine would use in casual everyday speech, whereas "día de pago" sounds like an unnatural, literal translation.
Round 22 of 118
The judge saw the two translations as A and B. The names in its reasons are filled in by us.
10 of the 16 ties are sentences where DeepL and AIRTranslate wrote exactly the same words, which is what a fair referee should call even.
DeepL offers one Spanish for all of Latin America.
Across Mexico, Argentina and Colombia, AIRTranslate was preferred 35 times out of 36.
02The examples
What winning looks like
Six rounds, word for word. The words in orange give DeepL's version away; the words in green are what a local says.
Spanish (Argentina)
“I'm broke until payday.”
- DeepL
- Estoy sin un peso hasta el día de pago.
- AIRTranslate
- Estoy sin un mango hasta que cobre.
An Argentine is broke sin un mango and gets paid when they cobran. El día de pago reads like a translation.
Spanish (Mexico)
“We're renting a small apartment with a pool.”
- DeepL
- Estamos alquilando un departamento pequeño con piscina.
- AIRTranslate
- Estamos rentando un depa chiquito con alberca.
In casual Mexican speech you rentas a depa with an alberca. DeepL's version is the textbook one.
French (Quebec)
“My girlfriend is coming over for dinner tonight.”
- DeepL
- Ma chum vient souper chez moi ce soir.
- AIRTranslate
- Ma blonde vient souper chez nous ce soir.
In Quebec your girlfriend is your blonde. Ma chum usually means a friend.
Arabic (Egypt)
“Give me a kilo of tomatoes and some bread.”
- DeepL
- أعطني كيلو من الطماطم وبعض الخبز.
- AIRTranslate
- هاتلي كيلو طماطم وشوية عيش.
In a Cairo market bread is عيش and you ask with هاتلي. DeepL writes formal Modern Standard Arabic.
Chinese (Taiwan)
“I'll take a taxi to the subway station.”
- DeepL
- 我要搭計程車去地鐵站。
- AIRTranslate
- 我會搭計程車去捷運站。
Taipei rides the 捷運. 地鐵 is the mainland word.
French (Quebec)
“It's freezing outside, wear a warm hat.”
- DeepL
- Il fait un froid de canard dehors, mets un tuque bien chaud.
- AIRTranslate
- Il fait frette dehors, mets-toi une tuque chaude.
Quebec says il fait frette, and tuque is feminine. DeepL used an idiom from France and got the gender wrong, in a variant it offers by name.
03DeepL's wins
The scoreboard is honest
DeepL won 3 rounds, and they matter, because they show the judge will rule against us. All 3 were small calls on wording and rhythm.
Spanish (Spain)
DeepL won this round“Hand me my sweater, it's getting cold.”
- DeepL
- Pásame el jersey, que empieza a hacer frío.
- AIRTranslate
- Pásame mi jersey, que está empezando a hacer frío.
In Spain you hand someone el jersey, not mi jersey. A fair call.
Portuguese (Brazil)
DeepL won this round“My phone battery died, can I borrow a charger?”
- DeepL
- A bateria do meu celular acabou, posso pegar um carregador emprestado?
- AIRTranslate
- Meu celular ficou sem bateria. Posso pegar um carregador emprestado?
The judge preferred a bateria acabou to our ficou sem bateria, while calling both great.
Chinese (Taiwan)
DeepL won this round“I bought fresh pineapple at the market.”
- DeepL
- 我在市場買了新鮮的鳳梨。
- AIRTranslate
- 我在市場買了新鮮鳳梨。
The judge wanted a 的 between 新鮮 and 鳳梨 for the rhythm of everyday speech.
04Availability
A continent is not a country
DeepL's answer to Latin America is one Spanish for the whole region, and it has one Arabic, written in the formal style. Egypt and Saudi Arabia, whose street speech splits in the first three words of a price haggle, finished 10 to 0 and 10 to 0. AIRTranslate offers Spanish for 23 countries and Arabic for 23, because Buenos Aires is not Mexico City and Cairo is not Riyadh.
474
AIRTranslate
country versions
Spanish for 23 countries, Arabic for 23, French for 32, English for 34, and more.
8
DeepL
country versions
US and UK English, Brazil and Portugal Portuguese, two Chinese scripts, Quebec French and one Latin American Spanish. One Arabic.
4
Google Translate
country versions
Quebec French, Portugal Portuguese and two Chinese scripts. One Spanish, one Arabic, for every country.
4
Apple Translate
country versions
US and UK English and two Chinese scripts. No country versions of Spanish, Arabic, French or Portuguese.
05Method
How the test works
- 1118
52 sentences, 11 countries
Everyday English, written so that a natural rendering differs by country in vocabulary, idiom, register or script. Sibling variants share the same sentences: 118 rounds in all.
- 2ES-419
DeepL at its strongest
Latin American Spanish for every Latin American variant, Quebec French for Quebec, both Portuguese variants natively, Traditional Chinese for Taiwan, and less formal wording to match the casual sentences.
- 3A / B
A blind referee from Google
Gemini 4 Argon, at its default settings, saw the two answers as A and B and was asked which a native of that country would find natural. If it leans toward anyone, it is not us.
- 4114/118
Twice, with the order swapped
A win counts only when both orders agree; anything else is a tie. 114 of 118 verdicts were the same both ways.
06The data
Check every call yourself
Every sentence, every translation from both systems and every verdict with its reason, so anyone can re-run the study or dispute a call.