We send exactly the same task prompt and the same materials to the 4 AIs on the same day. We open a new chat every time so earlier answers don't carry over. The sending order is rotated in each run. We also record the seconds from sending until the answer stops.

Why do you send the same task 3 times?

Because AI answers vary even for the same question. From a single try, you can't tell whether a result reflects real ability or chance. We send each task 3 times, and the share of tasks whose conclusion matched is reported as the consistency rate.

Who does the scoring?

Scoring is done only by a program. We do not let AI score the answers. The correct answers are set when the tasks are created, and whether an answer is right is decided by whether numbers match, whether required words are present, whether the format fits, whether URLs actually open and the result of running formulas. Matters of taste are not scored.

Is the data in the tasks from real companies?

No. We create new data for a fictional company every time. We do not use real companies, real people or text from other companies. We share the materials we used as they are, so you can repeat the same test.

How do you count cases where the AI couldn't answer?

Refusals and timeouts stay in the base and are counted as wrong. If we hit the plan's usage limit or a human verification screen appears, we remove that trial from the base and treat it as not measured. Operating errors on AInformation's side are also removed from the base, and we do not report them as the AI's fault.

Do you use the free or paid versions?

We use each AI's paid plan through the browser interface. We don't use the API. For each run, we record the plan name, the model name shown on screen and the time we checked it.

Does this ranking show which AI is the smartest?

No. We measure only 5 things on a fixed set of work tasks: accuracy, time to answer, consistency rate, real-source rate and instruction compliance. We do not measure writing style, readability or behavior in long conversations. If the tasks change, the ranking changes too.

Writing an apology email for a delayed delivery (4 AIs, 6 tasks, 3 runs each)

Name: Writing an apology email for a delayed delivery (4 AIs, 6 tasks, 3 runs each)
Creator: AInformation
Published: 2026-10-14
License: https://creativecommons.org/licenses/by/4.0/

RESULT

What this run found

The line below is generated directly from the scoring results. The page adds no words of its own.

Email and business documents: 6 tasks, 3 trials each Out of 6 tasks asked 3 times each, ChatGPT and Copilot tied for the highest accuracy rate at 100.0% 72 valid trials. Results from a single round of testing. Rankings may change on a different day or with different tasks. See scores by AI

SCORES

Scores by AI

Sorted by accuracy, highest first. The base is the number of trials in which that metric could be measured. “—” means not measured; it does not mean 0%.

Writing an apology email for a delayed delivery (4 AIs, 6 tasks, 3 runs each): scores by AI
AI	AccuracyCorrect ÷ valid trials	ConsistencySame conclusion in all 3 trials	Real-source rateShare of URLs that opened	Instruction complianceMet length and format rules	Time to answerMedian / range
1ChatGPT	100% 18 of 18 trials	100% out of 6 tasks	—	100%	17.3 s 14.1–23.7 seconds
2Copilot	100% 18 of 18 trials	100% out of 6 tasks	—	83.3%	17.1 s 14.1–20.3 seconds
3Claude	83.3% 15 of 18 trials	88.9% out of 6 tasks	—	50%	18.7 s 13.9–57.1 seconds
4Gemini	38.9% 7 of 18 trials	75% out of 6 tasks	—	90%	98.6 s 19.1–234 seconds

The real-source rate is measured only in runs that included a task asking for URLs along with the answer. For ChatGPT, Copilot, Claude and Gemini, “—” means this run did not include that kind of task. It does not mean 0%.

Trials with no answer
Gemini … Refused: 4 times / Timed out: 4 times
Refusals and timeouts stay in the base and count as incorrect. Plan limits, human verification screens and operating errors on AInformation's side are excluded from the base.

TASKS

Results by task

We sent each task to each of the 4 AIs 3 times. ○ means correct and × means incorrect. The numbers below are the seconds each trial took. Open each section to read the full task prompt, the answer key set in advance and why each verdict was given.

Easy

Based on the status memo below, write an apology email to send to the business partner. Rules - Output only the body of … Scoring type: contains_all — Whether all the required words and phrases are included

ChatGPT
- ChatGPT, trial 1: Correct 17.3 s
- ChatGPT, trial 2: Correct 17.3 s
- ChatGPT, trial 3: Correct 20.7 s
3 of 3 trials correct
Copilot
- Copilot, trial 1: Correct 20.2 s
- Copilot, trial 2: Correct 20.1 s
- Copilot, trial 3: Correct 20.3 s
3 of 3 trials correct
Claude
- Claude, trial 1: Correct 20.3 s
- Claude, trial 2: Correct 26.3 s
- Claude, trial 3: Correct 26.3 s
3 of 3 trials correct
Gemini
- Gemini, trial 1: Incorrect 19.1 s
- Gemini, trial 2: Incorrect 19.1 s
- Gemini, trial 3: Incorrect Refused
0 of 3 trials correct

Full task promptWe sent the original Japanese version of this prompt (427 Japanese characters) unchanged to all 4 AIs. The text shown here is our translation.

Based on the status memo below, write an apology email to send to the business partner. Rules - Output only the body of the email. Do not add an introduction or explanation - Do not use exclamation marks (!) - Include every date, quantity, and phone number from the memo without omitting any ---- Status memo starts here ---- [ Status memo for writing the email (fictional) ] To: Person in charge, Purchasing Department, Sample Bussan Co. From: Suzuki, Sales Department, Sample Trading Co. Phone: 03-4590-5977 - The customer ordered 650 units of Product C (cooling towel). - The original delivery date was October 27, 2026. - Because the cargo ship is stuck at port, delivery is now expected to shift to November 3, 2026. - The delay is 7 days. - Full delivery will be on November 3, but 325 units will be delivered early, on October 27. - No replacement product is available. Do not mention a discount this time. - The next update must be sent by October 17. ---- Status memo ends here ----

Answer key set in advanceWe set it when we created the task. 3 of the 4 AIs got it right at least once

(Automated scoring, contains_all) New delivery date = November 3 or 11/3 / Order quantity = 650 or 650 / Original delivery date = October 27 or 10/27 / Quantity delivered early = 325 or 325 / Apology wording = apology or sorry or regret / Phone number = 03-4590-5977

Scoring detailsWe show, trial by trial, why each verdict was given (9 of 12 trials correct)

ChatGPT, trial 1: Correct: All 6 items present (17.3 seconds) · Full text of this answer (Japanese)
ChatGPT, trial 2: Correct: All 6 items present (17.3 seconds) · Full text of this answer (Japanese)
ChatGPT, trial 3: Correct: All 6 items present (20.7 seconds) · Full text of this answer (Japanese)
Copilot, trial 1: Correct: All 6 items present (20.2 seconds) · Full text of this answer (Japanese)
Copilot, trial 2: Correct: All 6 items present (20.1 seconds) · Full text of this answer (Japanese)
Copilot, trial 3: Correct: All 6 items present (20.3 seconds) · Full text of this answer (Japanese)
Claude, trial 1: Correct: All 6 items present (20.3 seconds) · Full text of this answer (Japanese)
Claude, trial 2: Correct: All 6 items present (26.3 seconds) · Full text of this answer (Japanese)
Claude, trial 3: Correct: All 6 items present (26.3 seconds) · Full text of this answer (Japanese)
Gemini, trial 1: Incorrect: 0 of 6 items present - missing new delivery date / order quantity / original delivery date / quantity delivered early / apology wording / phone number (19.1 seconds) · Full text of this answer (Japanese)
Gemini, trial 2: Incorrect: 0 of 6 items present - missing new delivery date / order quantity / original delivery date / quantity delivered early / apology wording / phone number (19.1 seconds) · Full text of this answer (Japanese)
Gemini, trial 3: Incorrect: Refused (policy). Counted as incorrect and included in the denominator (22.4 seconds) · Full text of this answer (Japanese)

Easy

From the status memo below, write out three numbers. Rules - The answer must be exactly 3 lines, in this order: Order qu… Scoring type: exact_number — Whether the number matches the correct answer. The rounding tolerance is set in advance

ChatGPT
- ChatGPT, trial 1: Correct 14.3 s
- ChatGPT, trial 2: Correct 14.3 s
- ChatGPT, trial 3: Correct 14.1 s
3 of 3 trials correct
Copilot
- Copilot, trial 1: Correct 17.1 s
- Copilot, trial 2: Correct 17 s
- Copilot, trial 3: Correct 17.1 s
3 of 3 trials correct
Claude
- Claude, trial 1: Correct 14.9 s
- Claude, trial 2: Correct 14.1 s
- Claude, trial 3: Correct 13.9 s
3 of 3 trials correct
Gemini
- Gemini, trial 1: Correct 139.1 s
- Gemini, trial 2: Correct 37.3 s
- Gemini, trial 3: Correct 120 s
3 of 3 trials correct

Full task promptWe sent the original Japanese version of this prompt (414 Japanese characters) unchanged to all 4 AIs. The text shown here is our translation.

From the status memo below, write out three numbers. Rules - The answer must be exactly 3 lines, in this order: Order quantity: ○ / Quantity delivered early: ○ / Days delayed: ○ - Write numbers using half-width digits. Do not add an explanation ---- Status memo starts here ---- [ Status memo for writing the email (fictional) ] To: Person in charge, Purchasing Department, Sample Bussan Co. From: Suzuki, Sales Department, Sample Trading Co. Phone: 03-4590-5977 - The customer ordered 650 units of Product C (cooling towel). - The original delivery date was October 27, 2026. - Because the cargo ship is stuck at port, delivery is now expected to shift to November 3, 2026. - The delay is 7 days. - Full delivery will be on November 3, but 325 units will be delivered early, on October 27. - No replacement product is available. Do not mention a discount this time. - The next update must be sent by October 17. ---- Status memo ends here ----

Answer key set in advanceWe set it when we created the task. 4 of the 4 AIs got it right at least once

(Automated scoring, exact_number) Order quantity = 650 (tolerance ±0) / Quantity delivered early = 325 (tolerance ±0) / Days delayed = 7 (tolerance ±0)

Scoring detailsWe show, trial by trial, why each verdict was given (12 of 12 trials correct)

ChatGPT, trial 1: Correct: All numbers match: order quantity ○ (correct answer 650) / quantity delivered early ○ (correct answer 325) / days delayed ○ (correct answer 7) (14.3 seconds) · Full text of this answer (Japanese)
ChatGPT, trial 2: Correct: All numbers match: order quantity ○ (correct answer 650) / quantity delivered early ○ (correct answer 325) / days delayed ○ (correct answer 7) (14.3 seconds) · Full text of this answer (Japanese)
ChatGPT, trial 3: Correct: All numbers match: order quantity ○ (correct answer 650) / quantity delivered early ○ (correct answer 325) / days delayed ○ (correct answer 7) (14.1 seconds) · Full text of this answer (Japanese)
Copilot, trial 1: Correct: All numbers match: order quantity ○ (correct answer 650) / quantity delivered early ○ (correct answer 325) / days delayed ○ (correct answer 7) (17.1 seconds) · Full text of this answer (Japanese)
Copilot, trial 2: Correct: All numbers match: order quantity ○ (correct answer 650) / quantity delivered early ○ (correct answer 325) / days delayed ○ (correct answer 7) (17 seconds) · Full text of this answer (Japanese)
Copilot, trial 3: Correct: All numbers match: order quantity ○ (correct answer 650) / quantity delivered early ○ (correct answer 325) / days delayed ○ (correct answer 7) (17.1 seconds) · Full text of this answer (Japanese)
Claude, trial 1: Correct: All numbers match: order quantity ○ (correct answer 650) / quantity delivered early ○ (correct answer 325) / days delayed ○ (correct answer 7) (14.9 seconds) · Full text of this answer (Japanese)
Claude, trial 2: Correct: All numbers match: order quantity ○ (correct answer 650) / quantity delivered early ○ (correct answer 325) / days delayed ○ (correct answer 7) (14.1 seconds) · Full text of this answer (Japanese)
Claude, trial 3: Correct: All numbers match: order quantity ○ (correct answer 650) / quantity delivered early ○ (correct answer 325) / days delayed ○ (correct answer 7) (13.9 seconds) · Full text of this answer (Japanese)
Gemini, trial 1: Correct: All numbers match: order quantity ○ (correct answer 650) / quantity delivered early ○ (correct answer 325) / days delayed ○ (correct answer 7) (139.1 seconds) · Full text of this answer (Japanese)
Gemini, trial 2: Correct: All numbers match: order quantity ○ (correct answer 650) / quantity delivered early ○ (correct answer 325) / days delayed ○ (correct answer 7) (37.3 seconds) · Full text of this answer (Japanese)
Gemini, trial 3: Correct: All numbers match: order quantity ○ (correct answer 650) / quantity delivered early ○ (correct answer 325) / days delayed ○ (correct answer 7) (120 seconds) · Full text of this answer (Japanese)

Medium

Based on the status memo below, write the body of an apology email to send to the business partner, in 200 characters or… Scoring type: contains_all — Whether all the required words and phrases are included

ChatGPT
- ChatGPT, trial 1: Correct 14.2 s
- ChatGPT, trial 2: Correct 17.3 s
- ChatGPT, trial 3: Correct 17.4 s
3 of 3 trials correct
Copilot
- Copilot, trial 1: Correct 17.3 s
- Copilot, trial 2: Correct 17.1 s
- Copilot, trial 3: Correct 17.1 s
3 of 3 trials correct
Claude
- Claude, trial 1: Correct 14 s
- Claude, trial 2: Correct 17.1 s
- Claude, trial 3: Correct 17 s
3 of 3 trials correct
Gemini
- Gemini, trial 1: Incorrect Timed out
- Gemini, trial 2: Incorrect Refused
- Gemini, trial 3: Correct 215 s
1 of 3 trials correct

Full task promptWe sent the original Japanese version of this prompt (429 Japanese characters) unchanged to all 4 AIs. The text shown here is our translation.

Based on the status memo below, write the body of an apology email to send to the business partner, in 200 characters or fewer. Rules - Output only the body. Do not write a subject line or signature - 200 characters or fewer. Be sure to include the new delivery date, the quantity delivered early, and the number of days delayed ---- Status memo starts here ---- [ Status memo for writing the email (fictional) ] To: Person in charge, Purchasing Department, Sample Bussan Co. From: Suzuki, Sales Department, Sample Trading Co. Phone: 03-4590-5977 - The customer ordered 650 units of Product C (cooling towel). - The original delivery date was October 27, 2026. - Because the cargo ship is stuck at port, delivery is now expected to shift to November 3, 2026. - The delay is 7 days. - Full delivery will be on November 3, but 325 units will be delivered early, on October 27. - No replacement product is available. Do not mention a discount this time. - The next update must be sent by October 17. ---- Status memo ends here ----

Answer key set in advanceWe set it when we created the task. 4 of the 4 AIs got it right at least once

(Automated scoring, contains_all) New delivery date = November 3 or 11/3 / Quantity delivered early = 325 or 325 / Days delayed = 7 or 7 days

Scoring detailsWe show, trial by trial, why each verdict was given (10 of 12 trials correct)

ChatGPT, trial 1: Correct: All three items present (14.2 seconds) · Full text of this answer (Japanese)
ChatGPT, trial 2: Correct: All three items present (17.3 seconds) · Full text of this answer (Japanese)
ChatGPT, trial 3: Correct: All three items present (17.4 seconds) · Full text of this answer (Japanese)
Copilot, trial 1: Correct: All three items present (17.3 seconds) · Full text of this answer (Japanese)
Copilot, trial 2: Correct: All three items present (17.1 seconds) · Full text of this answer (Japanese)
Copilot, trial 3: Correct: All three items present (17.1 seconds) · Full text of this answer (Japanese)
Claude, trial 1: Correct: All three items present (14 seconds) · Full text of this answer (Japanese)
Claude, trial 2: Correct: All three items present (17.1 seconds) · Full text of this answer (Japanese)
Claude, trial 3: Correct: All three items present (17 seconds) · Full text of this answer (Japanese)
Gemini, trial 1: Incorrect: Timed out (the answer did not finish before the time limit) (302.7 seconds)
Gemini, trial 2: Incorrect: Refused (policy). Counted as incorrect and included in the denominator (19.4 seconds) · Full text of this answer (Japanese)
Gemini, trial 3: Correct: All three items present (215 seconds) · Full text of this answer (Japanese)

Medium

Based on the status memo below, write only the subject line of an apology email to send to the business partner. Rules -… Scoring type: format_check — Whether the answer followed the rules on character count, format and number of items

ChatGPT
- ChatGPT, trial 1: Correct 14.2 s
- ChatGPT, trial 2: Correct 14.3 s
- ChatGPT, trial 3: Correct 14.2 s
3 of 3 trials correct
Copilot
- Copilot, trial 1: Correct 16.9 s
- Copilot, trial 2: Correct 17.1 s
- Copilot, trial 3: Correct 14.1 s
3 of 3 trials correct
Claude
- Claude, trial 1: Incorrect 13.9 s
- Claude, trial 2: Incorrect 14 s
- Claude, trial 3: Incorrect 14 s
0 of 3 trials correct
Gemini
- Gemini, trial 1: Correct 77.2 s
- Gemini, trial 2: Incorrect Timed out
- Gemini, trial 3: Correct 208.7 s
2 of 3 trials correct

Full task promptWe sent the original Japanese version of this prompt (430 Japanese characters) unchanged to all 4 AIs. The text shown here is our translation.

Based on the status memo below, write only the subject line of an apology email to send to the business partner. Rules - The answer must be exactly 1 line. Write only the subject line text - 30 characters or fewer. Be sure to include the word “delivery” - Do not add “Subject:” or “Re:” ---- Status memo starts here ---- [ Status memo for writing the email (fictional) ] To: Person in charge, Purchasing Department, Sample Bussan Co. From: Suzuki, Sales Department, Sample Trading Co. Phone: 03-4590-5977 - The customer ordered 650 units of Product C (cooling towel). - The original delivery date was October 27, 2026. - Because the cargo ship is stuck at port, delivery is now expected to shift to November 3, 2026. - The delay is 7 days. - Full delivery will be on November 3, but 325 units will be delivered early, on October 27. - No replacement product is available. Do not mention a discount this time. - The next update must be sent by October 17. ---- Status memo ends here ----

Answer key set in advanceWe set it when we created the task. 3 of the 4 AIs got it right at least once

(Automated scoring, format_check) {'type': 'line_count', 'min': 1, 'max': 1} / {'type': 'max_chars', 'value': 30} / {'type': 'must_contain', 'values': ['delivery']} / {'type': 'must_not_contain', 'values': ['Re:', 'Subject']}

Scoring detailsWe show, trial by trial, why each verdict was given (8 of 12 trials correct)

ChatGPT, trial 1: Correct: All format checks pass: line_count ○ (1 line / 1–1 line) / max_chars ○ (12 characters / limit 30) / must_contain ○ (all present) / must_not_contain ○ (none present) (14.2 seconds) · Full text of this answer (Japanese)
ChatGPT, trial 2: Correct: All format checks pass: line_count ○ (1 line / 1–1 line) / max_chars ○ (24 characters / limit 30) / must_contain ○ (all present) / must_not_contain ○ (none present) (14.3 seconds) · Full text of this answer (Japanese)
ChatGPT, trial 3: Correct: All format checks pass: line_count ○ (1 line / 1–1 line) / max_chars ○ (19 characters / limit 30) / must_contain ○ (all present) / must_not_contain ○ (none present) (14.2 seconds) · Full text of this answer (Japanese)
Copilot, trial 1: Correct: All format checks pass: line_count ○ (1 line / 1–1 line) / max_chars ○ (19 characters / limit 30) / must_contain ○ (all present) / must_not_contain ○ (none present) (16.9 seconds) · Full text of this answer (Japanese)
Copilot, trial 2: Correct: All format checks pass: line_count ○ (1 line / 1–1 line) / max_chars ○ (17 characters / limit 30) / must_contain ○ (all present) / must_not_contain ○ (none present) (17.1 seconds) · Full text of this answer (Japanese)
Copilot, trial 3: Correct: All format checks pass: line_count ○ (1 line / 1–1 line) / max_chars ○ (17 characters / limit 30) / must_contain ○ (all present) / must_not_contain ○ (none present) (14.1 seconds) · Full text of this answer (Japanese)
Claude, trial 1: Incorrect: Format does not match: line_count × (7 lines / 1–1 line) / max_chars ○ (30 characters / limit 30) / must_contain ○ (all present) / must_not_contain ○ (none present) (13.9 seconds) · Full text of this answer (Japanese)
Claude, trial 2: Incorrect: Format does not match: line_count × (7 lines / 1–1 line) / max_chars ○ (23 characters / limit 30) / must_contain ○ (all present) / must_not_contain ○ (none present) (14 seconds) · Full text of this answer (Japanese)
Claude, trial 3: Incorrect: Format does not match: line_count × (8 lines / 1–1 line) / max_chars × (35 characters / limit 30) / must_contain ○ (all present) / must_not_contain ○ (none present) (14 seconds) · Full text of this answer (Japanese)
Gemini, trial 1: Correct: All format checks pass: line_count ○ (1 line / 1–1 line) / max_chars ○ (23 characters / limit 30) / must_contain ○ (all present) / must_not_contain ○ (none present) (77.2 seconds) · Full text of this answer (Japanese)
Gemini, trial 2: Incorrect: Timed out (the answer did not finish before the time limit) (302.9 seconds)
Gemini, trial 3: Correct: All format checks pass: line_count ○ (1 line / 1–1 line) / max_chars ○ (29 characters / limit 30) / must_contain ○ (all present) / must_not_contain ○ (none present) (208.7 seconds) · Full text of this answer (Japanese)

Hard

Based on the status memo below, write an apology email to send to the business partner. Rules - Address it to: Person in… Scoring type: contains_all — Whether all the required words and phrases are included

ChatGPT
- ChatGPT, trial 1: Correct 23.6 s
- ChatGPT, trial 2: Correct 23.7 s
- ChatGPT, trial 3: Correct 20.2 s
3 of 3 trials correct
Copilot
- Copilot, trial 1: Correct 20.1 s
- Copilot, trial 2: Correct 17.2 s
- Copilot, trial 3: Correct 17 s
3 of 3 trials correct
Claude
- Claude, trial 1: Correct 41.7 s
- Claude, trial 2: Correct 51.2 s
- Claude, trial 3: Correct 57.1 s
3 of 3 trials correct
Gemini
- Gemini, trial 1: Incorrect 28.2 s
- Gemini, trial 2: Correct 234 s
- Gemini, trial 3: Incorrect Refused
1 of 3 trials correct

Full task promptWe sent the original Japanese version of this prompt (466 Japanese characters) unchanged to all 4 AIs. The text shown here is our translation.

Based on the status memo below, write an apology email to send to the business partner. Rules - Address it to: Person in charge, Purchasing Department, Sample Bussan Co. - 400 characters or fewer overall - Do not use exclamation marks (!) - Include every date, quantity, and phone number from the memo without omitting any - Output only the body. Do not add an introduction or explanation ---- Status memo starts here ---- [ Status memo for writing the email (fictional) ] To: Person in charge, Purchasing Department, Sample Bussan Co. From: Suzuki, Sales Department, Sample Trading Co. Phone: 03-4590-5977 - The customer ordered 650 units of Product C (cooling towel). - The original delivery date was October 27, 2026. - Because the cargo ship is stuck at port, delivery is now expected to shift to November 3, 2026. - The delay is 7 days. - Full delivery will be on November 3, but 325 units will be delivered early, on October 27. - No replacement product is available. Do not mention a discount this time. - The next update must be sent by October 17. ---- Status memo ends here ----

Answer key set in advanceWe set it when we created the task. 4 of the 4 AIs got it right at least once

(Automated scoring, contains_all) New delivery date = November 3 or 11/3 / Order quantity = 650 or 650 / Original delivery date = October 27 or 10/27 / Quantity delivered early = 325 or 325 / Apology wording = apology or sorry or regret / Phone number = 03-4590-5977

Scoring detailsWe show, trial by trial, why each verdict was given (10 of 12 trials correct)

ChatGPT, trial 1: Correct: All 6 items present (23.6 seconds) · Full text of this answer (Japanese)
ChatGPT, trial 2: Correct: All 6 items present (23.7 seconds) · Full text of this answer (Japanese)
ChatGPT, trial 3: Correct: All 6 items present (20.2 seconds) · Full text of this answer (Japanese)
Copilot, trial 1: Correct: All 6 items present (20.1 seconds) · Full text of this answer (Japanese)
Copilot, trial 2: Correct: All 6 items present (17.2 seconds) · Full text of this answer (Japanese)
Copilot, trial 3: Correct: All 6 items present (17 seconds) · Full text of this answer (Japanese)
Claude, trial 1: Correct: All 6 items present (41.7 seconds) · Full text of this answer (Japanese)
Claude, trial 2: Correct: All 6 items present (51.2 seconds) · Full text of this answer (Japanese)
Claude, trial 3: Correct: All 6 items present (57.1 seconds) · Full text of this answer (Japanese)
Gemini, trial 1: Incorrect: 0 of 6 items present - missing new delivery date / order quantity / original delivery date / quantity delivered early / apology wording / phone number (28.2 seconds) · Full text of this answer (Japanese)
Gemini, trial 2: Correct: All 6 items present (234 seconds) · Full text of this answer (Japanese)
Gemini, trial 3: Incorrect: Refused (policy). Counted as incorrect and included in the denominator (144.4 seconds) · Full text of this answer (Japanese)

Hard

Based on the status memo below, write an apology email to send to the business partner, and output only JSON. Rules - Ou… Scoring type: contains_all — Whether all the required words and phrases are included

ChatGPT
- ChatGPT, trial 1: Correct 17.4 s
- ChatGPT, trial 2: Correct 20.7 s
- ChatGPT, trial 3: Correct 20.3 s
3 of 3 trials correct
Copilot
- Copilot, trial 1: Correct 20.1 s
- Copilot, trial 2: Correct 20.3 s
- Copilot, trial 3: Correct 20.1 s
3 of 3 trials correct
Claude
- Claude, trial 1: Correct 23.2 s
- Claude, trial 2: Correct 32.6 s
- Claude, trial 3: Correct 26.1 s
3 of 3 trials correct
Gemini
- Gemini, trial 1: Incorrect Refused
- Gemini, trial 2: Incorrect Timed out
- Gemini, trial 3: Incorrect Timed out
0 of 3 trials correct

Full task promptWe sent the original Japanese version of this prompt (470 Japanese characters) unchanged to all 4 AIs. The text shown here is our translation.

Based on the status memo below, write an apology email to send to the business partner, and output only JSON. Rules - Output only JSON. Do not write any explanatory text before or after it - Use the format {"subject": "Subject", "body": "Body"} - Include the dates, quantities, and phone number from the memo in the body without omitting any of them ---- Status memo starts here ---- [ Status memo for writing the email (fictional) ] To: Person in charge, Purchasing Department, Sample Bussan Co. From: Suzuki, Sales Department, Sample Trading Co. Phone: 03-4590-5977 - The customer ordered 650 units of Product C (cooling towel). - The original delivery date was October 27, 2026. - Because the cargo ship is stuck at port, delivery is now expected to shift to November 3, 2026. - The delay is 7 days. - Full delivery will be on November 3, but 325 units will be delivered early, on October 27. - No replacement product is available. Do not mention a discount this time. - The next update must be sent by October 17. ---- Status memo ends here ----

Answer key set in advanceWe set it when we created the task. 3 of the 4 AIs got it right at least once

(Automated scoring, contains_all) New delivery date = November 3 or 11/3 / Order quantity = 650 or 650 / Original delivery date = October 27 or 10/27 / Quantity delivered early = 325 or 325 / Apology wording = apology or sorry or regret / Phone number = 03-4590-5977

Scoring detailsWe show, trial by trial, why each verdict was given (9 of 12 trials correct)

ChatGPT, trial 1: Correct: All 6 items present (17.4 seconds) · Full text of this answer (Japanese)
ChatGPT, trial 2: Correct: All 6 items present (20.7 seconds) · Full text of this answer (Japanese)
ChatGPT, trial 3: Correct: All 6 items present (20.3 seconds) · Full text of this answer (Japanese)
Copilot, trial 1: Correct: All 6 items present (20.1 seconds) · Full text of this answer (Japanese)
Copilot, trial 2: Correct: All 6 items present (20.3 seconds) · Full text of this answer (Japanese)
Copilot, trial 3: Correct: All 6 items present (20.1 seconds) · Full text of this answer (Japanese)
Claude, trial 1: Correct: All 6 items present (23.2 seconds) · Full text of this answer (Japanese)
Claude, trial 2: Correct: All 6 items present (32.6 seconds) · Full text of this answer (Japanese)
Claude, trial 3: Correct: All 6 items present (26.1 seconds) · Full text of this answer (Japanese)
Gemini, trial 1: Incorrect: Refused (policy). Counted as incorrect and included in the denominator (37.4 seconds) · Full text of this answer (Japanese)
Gemini, trial 2: Incorrect: Timed out (the answer did not finish before the time limit) (301.9 seconds)
Gemini, trial 3: Incorrect: Timed out (the answer did not finish before the time limit) (301.7 seconds)

ANSWERS

Full answers from the 4 AIs

These are the answers we scored, posted as is without edits. That way, anyone can recount the results the same way we did.

ChatGPT

18 answers18 of them correct

t1, trial 1

Full text file (Japanese) · Correct

t1, trial 2

Full text file (Japanese) · Correct

t1, trial 3

Full text file (Japanese) · Correct

t2, trial 1

Full text file (Japanese) · Correct

t2, trial 2

Full text file (Japanese) · Correct

t2, trial 3

Full text file (Japanese) · Correct

t3, trial 1

Full text file (Japanese) · Correct

t3, trial 2

Full text file (Japanese) · Correct

t3, trial 3

Full text file (Japanese) · Correct

t4, trial 1

Full text file (Japanese) · Correct

t4, trial 2

Full text file (Japanese) · Correct

t4, trial 3

Full text file (Japanese) · Correct

t5, trial 1

Full text file (Japanese) · Correct

t5, trial 2

Full text file (Japanese) · Correct

t5, trial 3

Full text file (Japanese) · Correct

t6, trial 1

Full text file (Japanese) · Correct

t6, trial 2

Full text file (Japanese) · Correct

t6, trial 3

Full text file (Japanese) · Correct
Copilot

18 answers18 of them correct

t1, trial 1

Full text file (Japanese) · Correct

t1, trial 2

Full text file (Japanese) · Correct

t1, trial 3

Full text file (Japanese) · Correct

t2, trial 1

Full text file (Japanese) · Correct

t2, trial 2

Full text file (Japanese) · Correct

t2, trial 3

Full text file (Japanese) · Correct

t3, trial 1

Full text file (Japanese) · Correct

t3, trial 2

Full text file (Japanese) · Correct

t3, trial 3

Full text file (Japanese) · Correct

t4, trial 1

Full text file (Japanese) · Correct

t4, trial 2

Full text file (Japanese) · Correct

t4, trial 3

Full text file (Japanese) · Correct

t5, trial 1

Full text file (Japanese) · Correct

t5, trial 2

Full text file (Japanese) · Correct

t5, trial 3

Full text file (Japanese) · Correct

t6, trial 1

Full text file (Japanese) · Correct

t6, trial 2

Full text file (Japanese) · Correct

t6, trial 3

Full text file (Japanese) · Correct
Claude

18 answers15 of them correct

t1, trial 1

Full text file (Japanese) · Correct

t1, trial 2

Full text file (Japanese) · Correct

t1, trial 3

Full text file (Japanese) · Correct

t2, trial 1

Full text file (Japanese) · Correct

t2, trial 2

Full text file (Japanese) · Correct

t2, trial 3

Full text file (Japanese) · Correct

t3, trial 1

Full text file (Japanese) · Correct

t3, trial 2

Full text file (Japanese) · Correct

t3, trial 3

Full text file (Japanese) · Correct

t4, trial 1

Full text file (Japanese) · Incorrect

t4, trial 2

Full text file (Japanese) · Incorrect

t4, trial 3

Full text file (Japanese) · Incorrect

t5, trial 1

Full text file (Japanese) · Correct

t5, trial 2

Full text file (Japanese) · Correct

t5, trial 3

Full text file (Japanese) · Correct

t6, trial 1

Full text file (Japanese) · Correct

t6, trial 2

Full text file (Japanese) · Correct

t6, trial 3

Full text file (Japanese) · Correct
Gemini

14 answers7 of them correct

t1, trial 1

Full text file (Japanese) · Incorrect

t1, trial 2

Full text file (Japanese) · Incorrect

t1, trial 3

Full text file (Japanese) · Incorrect

t2, trial 1

Full text file (Japanese) · Correct

t2, trial 2

Full text file (Japanese) · Correct

t2, trial 3

Full text file (Japanese) · Correct

t3, trial 2

Full text file (Japanese) · Incorrect

t3, trial 3

Full text file (Japanese) · Correct

t4, trial 1

Full text file (Japanese) · Correct

t4, trial 3

Full text file (Japanese) · Correct

t5, trial 1

Full text file (Japanese) · Incorrect

t5, trial 2

Full text file (Japanese) · Correct

t5, trial 3

Full text file (Japanese) · Incorrect

t6, trial 1

Full text file (Japanese) · Incorrect

MATERIALS

Materials used in the tasks

A "delivery is delayed" status memo for a fictional company, Sample Trading Co. The elements it must include were decided when it was created We recreated them for this run on October 13, 2026. Download them to repeat the same test yourself.

mail_brief.txt 692B (Japanese)

METHOD

How we measure

We list this run's conditions exactly as they were, so anyone can repeat it.

AIs measured and their model names at the time

AI services change under the hood without notice. We record the model name shown on screen and the time we checked it.

AIs measured and model names
AI	Plan	Model name shown on screen	Time checked
ChatGPT	ChatGPT Pro	Medium	2026-10-13 20:17
Gemini	Google AI Pro	Flash (high cost)	2026-10-13 20:17
Claude	Claude Max	Opus 4.8 high	2026-10-13 20:17
Copilot	Microsoft 365 Personal		2026-10-13 20:17

What we did in this run

We recreated the materials for the tasks specifically for this run. The data are for a fictional company and include no real companies or people.
We created 6 tasks and set the answer keys at that point.
We sent the same prompt and the same materials to ChatGPT, Gemini, Claude and Copilot.
We sent each task 3 times, in a new chat each time.
We rotated the sending order from run to run. This keeps the time of day from skewing the ranking.
We waited up to 300 seconds for each answer, and recorded anything longer as a timeout.
Scoring: Machine scoring only (not graded by AI).
We sent 72 trials in total, and 72 of them were valid.

How we count trials with no answer

Refused or timed out
We keep it in the base and count it as wrong. Because at work, that is the same as getting no answer.
Hit the plan's usage limit
We remove it from the base. We treat that trial as not measured.
A human verification screen appeared
We stop there. We neither click through it automatically nor try to get around it.
Operating error on AInformation's side
We remove it from the base. We do not report it as the AI's fault.

LIMITS

Limits of this benchmark

So that no one judges by the numbers alone, we list what we could not measure.

This is a single run on a single day. Rankings can change with the day or the task
The plan used is each AI's paid tier (ChatGPT Pro / Google AI Pro / Claude Max / Microsoft 365 Personal). These are not results from the free tier
Measurements use only the default model and default mode (features such as Deep Research are not used)
The tasks use original data for the fictional company Sample Shoji Co., Ltd. These are not results using data from a real company
Only tasks that can be scored by machine are covered. Writing quality and readability are not measured
Measurements are taken from the same screen a person would use in a browser. These are not results from the API
For any AI whose on-screen model name could not be read in a given run, that field is left blank (null). When the name could be read, it is shown as is

DATA

Published data

In the CSV, 1 row is 1 trial. The figures on this page match a recount of that CSV. The license is CC BY 4.0.

Latest version (CSV) (Japanese) 1 row = 1 trial, 14KB
Latest version (JSON) 42KB
v1 (version at publication) (CSV) (Japanese) 1 row = 1 trial, 14KB
v1 (version at publication) (JSON) 42KB

The cumulative rankings are at /en/data/research/bench/leaderboard.json. We do not delete versions (such as v1). That way, links from people who cited them keep working.

FAQ

How do you measure?: We send exactly the same task prompt and the same materials to the 4 AIs on the same day. We open a new chat every time so earlier answers don't carry over. The sending order is rotated in each run. We also record the seconds from sending until the answer stops.
Why do you send the same task 3 times?: Because AI answers vary even for the same question. From a single try, you can't tell whether a result reflects real ability or chance. We send each task 3 times, and the share of tasks whose conclusion matched is reported as the consistency rate.
Who does the scoring?: Scoring is done only by a program. We do not let AI score the answers. The correct answers are set when the tasks are created, and whether an answer is right is decided by whether numbers match, whether required words are present, whether the format fits, whether URLs actually open and the result of running formulas. Matters of taste are not scored.
Is the data in the tasks from real companies?: No. We create new data for a fictional company every time. We do not use real companies, real people or text from other companies. We share the materials we used as they are, so you can repeat the same test.
How do you count cases where the AI couldn't answer?: Refusals and timeouts stay in the base and are counted as wrong. If we hit the plan's usage limit or a human verification screen appears, we remove that trial from the base and treat it as not measured. Operating errors on AInformation's side are also removed from the base, and we do not report them as the AI's fault.
Do you use the free or paid versions?: We use each AI's paid plan through the browser interface. We don't use the API. For each run, we record the plan name, the model name shown on screen and the time we checked it.
Does this ranking show which AI is the smartest?: No. We measure only 5 things on a fixed set of work tasks: accuracy, time to answer, consistency rate, real-source rate and instruction compliance. We do not measure writing style, readability or behavior in long conversations. If the tasks change, the ranking changes too.
Can I cite these results?: Yes. The license is CC BY 4.0. As the source, please write “Source: AInformation, AI Practical Work Benchmark” with the URL of the page you used and the measurement date. The results are also available as JSON and CSV. With the CSV (1 row per trial), you can recount the figures we publish.

CITATION

Cite this benchmark

Source: AInformation, “AI Practical Work Benchmark: Writing an apology email for a delayed delivery (4 AIs, 6 tasks, 3 runs each)” (https://ainformation.jp/en/ai-surveys/ai-work-benchmark/apology-email-2026-10-14). Measured on October 14, 2026. 6 tasks, 3 trials each, 72 valid trials

When you cite a figure, please include the base too (how many of how many trials). Please cite AInformation as the source, along with the URL of the page you used and the measurement date.

What this run found

Scores by AI

Results by task

Full answers from the 4 AIs

Materials used in the tasks

How we measure

AIs measured and their model names at the time

What we did in this run

How we count trials with no answer

Limits of this benchmark

Published data

FAQ

Cite this benchmark

Related pages