Skip to content
Placement Papers - Free Placement Papers for Top CompaniesPlacement Papers - Free Placement Papers for Top CompaniesPlacement Papers

Capgemini Previous Year Question Papers PDF and Exam Pattern 2026

Capgemini previous year question papers are practice sets that copy the reported Capgemini online test: a pseudocode tracing section, quantitative aptitude, game-based aptitude, English or technical MCQs, and a separate Coding Benchmark. The 2026 pattern students report is a multi-section Stage 1 followed by a Coding Benchmark of 2 problems in 45 minutes. Older packs put pseudocode at about 15 MCQs in 25 minutes. This page covers 2024, 2025 and 2026 with worked questions, and the full 2026 set is a free PDF.

Capgemini exam pattern by role / drive

The numbers below are candidate-reported. Capgemini changes section order and clocks between campus, Exceller off-campus and other drives. Your invite is the final word.

Drive / year Section Questions Time Notes
2024 campus packs Game-based pseudocode 15 MCQs 25 min Steep filter. About 35% cleared it in student notes
2024 campus packs Hands-on coding 2 problems 45 min Easy-medium string and number problems
2025 drives Game-based pseudocode 15 MCQs 25 min Adaptive difficulty. More recursion and stack tracing
2025 drives Hands-on coding 2 problems 45 min Partial marking on some drives
2026 Stage 1 Pseudocode Varies by drive Varies by drive Language-agnostic tracing, no IDE
2026 Stage 1 Quantitative Varies by drive Varies by drive Percentages, profit and loss, time and work, ratios, DI
2026 Stage 1 Game-Based Aptitude Often 4 mini-games About 20-30 min Attention, memory, patterns, deduction
2026 Stage 1 English / tech MCQs Varies by drive Varies by drive Grammar, RC, programming, cloud, DBMS, OS, networks
2026 Coding Benchmark Coding 2 problems 45 min C, C++, Java. Python reported blocked on several invites

On Exceller drives, Coding Benchmark solves set your band once Stage 1 is cleared: 0 solves maps to Analyst, 1 to Analyst Star, 2 to Senior Analyst (candidate-reported). Most reports say there is no negative marking.

Year-wise question sets

The questions below are new, written in each year’s reported style, so they add volume on top of the year pages.

Capgemini 2026 question paper (Stage 1 + Coding Benchmark)

2026 is the year the test became clearly multi-section. Stage 1 now mixes pseudocode, quant, game-based aptitude and English or tech MCQs, and the Coding Benchmark runs as its own stage. Tech MCQs pick up cloud and AI vocabulary, and several Coding Benchmark invites blocked Python.

Q: Pseudocode. What does this print? SET x = 10, y = 3. WHILE x > y: SET x = x - y, then SET y = y + 1. END WHILE. PRINT x, y

Trace each pass:

  • Start: x = 10, y = 3. 10 > 3, so x = 7, y = 4.
  • 7 > 4, so x = 3, y = 5.
  • 3 > 5 is false. Loop ends.

Answer: 3 5

Q: Quant. A shopkeeper marks an item 40% above cost and then gives a 15% discount on the marked price. What is his profit percentage?

Take cost = 100. Marked price = 140. Selling price = 140 × 0.85 = 119.

Profit = 119 - 100 = 19 on a cost of 100.

Answer: 19%

Q: English. Find the error: “Each of the candidates have submitted their forms before the deadline.”

“Each” is singular, so the verb must be singular. “Have submitted” should be “has submitted”.

Answer: “have” should be “has”

Q: Tech MCQ. Which cloud service model lets you deploy your code on a managed runtime without managing servers or the operating system? (a) IaaS (b) PaaS (c) SaaS (d) On-premise

IaaS gives you virtual machines, so you still manage the OS. SaaS is a finished application you only use. PaaS gives you a managed runtime where you only push code and configuration.

Answer: (b) PaaS

Q: Coding Benchmark. Given n integers, print the second largest distinct value, or -1 if it does not exist. Example: [5, 9, 9, 3] gives 5.

Keep the largest and second largest in one pass. Skip values equal to the current largest so duplicates do not count. Java, since Python may be blocked:

import java.util.*;


public class Main {
    public static void main(String[] args) {
        Scanner sc = new Scanner(System.in);
        int n = sc.nextInt();
        long first = Long.MIN_VALUE, second = Long.MIN_VALUE;
        for (int i = 0; i < n; i++) {
            long x = sc.nextLong();
            if (x > first) { second = first; first = x; }
            else if (x < first && x > second) { second = x; }
        }
        System.out.println(second == Long.MIN_VALUE ? -1 : second);
    }
}

Answer: one pass, O(n) time and O(1) space. For [5, 9, 9, 3] it prints 5. For [4, 4] it prints -1.

Full 2026 notes and plan: Capgemini 2026 pattern details.

Capgemini 2025 question paper (adaptive pseudocode)

In 2025 the game-based pseudocode block turned adaptive, so a correct answer pulled a harder next question. Tracing moved toward recursion and data structures: recursive sums, binary search, 2D arrays, stacks, GCD and sort passes. Coding got partial marking on some drives.

Q: Pseudocode. FUNCTION f(n): IF n <= 1 RETURN 1, ELSE RETURN f(n - 1) + 2 × f(n - 2). What is f(5)?

Build up from the base case instead of expanding the tree:

  • f(0) = 1, f(1) = 1
  • f(2) = f(1) + 2 × f(0) = 1 + 2 = 3
  • f(3) = 3 + 2 × 1 = 5
  • f(4) = 5 + 2 × 3 = 11
  • f(5) = 11 + 2 × 5 = 21

Answer: 21

Q: Stack. Starting empty: PUSH 4, PUSH 7, POP, PUSH 2, PUSH 9, POP, PUSH 5. What is the top element and the stack size?
  • PUSH 4 → [4]
  • PUSH 7 → [4, 7]
  • POP removes 7 → [4]
  • PUSH 2, PUSH 9 → [4, 2, 9]
  • POP removes 9 → [4, 2]
  • PUSH 5 → [4, 2, 5]

Answer: top = 5, size = 3

Q: Binary search on a sorted array of 31 elements. What is the maximum number of comparisons needed to find an element or decide it is absent?

Each comparison halves the search range. The worst case is floor(log2 n) + 1. For n = 31, log2 31 is just under 5, so floor is 4 and the count is 5.

Answer: 5

Q: Quant. A can finish a job in 12 days and B in 18 days. They work together for 4 days, then A leaves. How many more days does B need?

Combined rate = 1/12 + 1/18 = 3/36 + 2/36 = 5/36 per day.

In 4 days they finish 20/36 = 5/9. Remaining = 4/9.

B alone: (4/9) × 18 = 8 days.

Answer: 8 days

Q: Coding. Move all zeros in an array to the end while keeping the order of the non-zero elements. Example: [0, 3, 0, 1, 7] becomes [3, 1, 7, 0, 0].

Use a write pointer. Copy every non-zero forward, then fill the rest with zeros. C++:

#include <bits/stdc++.h>
using namespace std;


int main() {
    int n; cin >> n;
    vector<int> a(n);
    for (auto &x : a) cin >> x;
    int w = 0;
    for (int i = 0; i < n; i++)
        if (a[i] != 0) a[w++] = a[i];
    while (w < n) a[w++] = 0;
    for (int x : a) cout << x << " ";
    return 0;
}

Answer: O(n) time, O(1) extra space, stable for non-zero values. With partial marking, submit a working version early, then harden it.

Full 2025 set: Capgemini 2025 papers.

Capgemini 2024 question paper (pseudocode + hands-on coding)

The 2024 packs describe a 15-question pseudocode block in 25 minutes, followed by 2 coding problems in 45 minutes. Tracing themes were basic: loop output, sums, conditionals, nested loops, string length, factorial and prime checks. Coding stayed with reverse, palindrome, Armstrong and Fibonacci style problems.

Q: Pseudocode. SET c = 0. FOR i = 1 TO 4: FOR j = i TO 4: SET c = c + 1. PRINT c

The inner loop runs 4 times for i = 1, then 3, 2 and 1 times.

4 + 3 + 2 + 1 = 10.

Answer: 10

Q: Pseudocode. SET n = 2024, s = 0. WHILE n > 0: SET s = s + (n MOD 10), SET n = n / 10 (integer division). PRINT s

The loop peels off the last digit each time: 4, 2, 0, 2.

s = 4 + 2 + 0 + 2 = 8.

Answer: 8 (the digit sum)

Q: Pseudocode. SET a = 5, b = 0. IF b != 0 AND a / b > 1 THEN PRINT “X” ELSE PRINT “Y”. What happens?

AND short-circuits. Since b != 0 is false, the right side a / b is never evaluated, so there is no division-by-zero error. The ELSE branch runs.

Answer: prints Y

Q: Quant. A price rises by 20% and then falls by 20%. What is the net change?

100 → 120 → 120 × 0.8 = 96. Net change = -4.

Answer: 4% decrease

Q: Coding. Check whether a number is a strong number (the sum of the factorials of its digits equals the number). Example: 145 = 1! + 4! + 5!.

Precompute factorials 0! to 9!, then add them digit by digit. Python (allowed on 2024 drives):

n = int(input())
fact = [1] * 10
for i in range(1, 10):
    fact[i] = fact[i - 1] * i
total, t = 0, n
while t > 0:
    total += fact[t % 10]
    t //= 10
print("Strong" if total == n else "Not strong")

Answer: 1 + 24 + 120 = 145, so 145 is strong. O(d) time for d digits. Watch the edge case n = 0: the loop never runs, so handle it on its own if the problem counts 0! = 1.

Full 2024 set: Capgemini 2024 papers.

Section-wise topics that repeat

Capgemini papers repeat themes, not exact questions, so a 2024 set still trains you for the 2026 Stage 1.

Section Topics that keep coming back How many Qs
Pseudocode Loop output, array sums and reverse, nested loops with break/continue, short-circuit conditionals, off-by-one indexing. From 2025: recursion with a tricky base case, stack and binary search counting, GCD, bubble-sort passes About 15 in older packs. Varies in 2026
Quantitative Percentages, profit and loss, time and work, ratios, data interpretation Varies by drive
Game-Based Aptitude Attention, memory, pattern and deduction mini-games Often 4 games in 2026
English / tech MCQs Grammar, RC, short writing. Programming basics, cloud models, DBMS, OS, networks Varies by drive
Coding Reverse, palindrome, Armstrong, Fibonacci, vowel count (2024). Rotate array, two sum, remove duplicates (2025). Easy-medium string and array work (2026) 2 problems

How to use one paper in a timed sitting

  1. Set the pseudocode clock first. On the older 15-question format that is 25 minutes, about 1.5 minutes per question. Skip any trace that is still stuck after 90 to 120 seconds.
  2. Trace on paper, not in an IDE. Write each variable’s value line by line. Mark every question you guessed.
  3. Run quant and English or tech MCQs on the clock your invite states. Do not leave any MCQ blank, since most reports show no negative marking.
  4. If the paper has a game module, play real attention and memory drills for about 20 to 30 minutes. Reading about games does not build the skill.
  5. Give coding a separate 45-minute block for 2 problems. Skim both, finish the sure one in about 20 minutes, then start the second. Use C++ or Java if your drive blocks Python.
  6. Score it. Then spend 15 minutes listing which tracing idea or edge case you missed. Three reviewed papers beat ten skimmed ones.

FAQs

Is there a Capgemini previous year question paper PDF?

Yes. The free Capgemini 2026 placement paper PDF is here, and the 2024, 2025 and 2026 year pages have more solved questions. These are practice sets rebuilt from candidate reports in the reported pattern, not official or leaked papers.

What is the Capgemini exam pattern for 2026?

Candidate-reported: a multi-section Stage 1 (Pseudocode, Quantitative, Game-Based Aptitude with often 4 mini-games in about 20-30 minutes, and English or technical MCQs), then a separate Coding Benchmark of 2 problems in 45 minutes, then technical and HR interviews. Section order and clocks vary by drive, so trust your invite.

Is there negative marking in the Capgemini online test?

Most candidate reports say there is no negative marking, so attempt every question. Confirm it on the instructions screen of your drive.

Which Capgemini paper year should I start with?

Start with the 2024 set to build pseudocode tracing speed on the 15 questions in 25 minutes format, move to 2025 for recursion and stack tracing, and finish with the 2026 PDF as a full timed run because it matches the current multi-section Stage 1.

Is Python allowed in the Capgemini Coding Benchmark?

Several 2026 Coding Benchmark invites reported Python as blocked, with C, C++ and Java allowed. Older 2024 and 2025 reports still list Python. Read your mailer and practise in C++ or Java if it is not stated.

Comments & Suggestions