CAT, XAT, NMAT, SNAP, CMAT and MAT compared

Six MBA entrance tests, six different papers: MAT allows 48 seconds a question, CMAT 108, and CAT does not publish how many questions it has at all. A wrong answer costs a third of a right answer’s marks on a CAT MCQ, a quarter in XAT, SNAP and CMAT, and nothing in NMAT.

This lesson puts the six side by side from each test’s own documents, never from second-hand tables. A program works out the time each pattern allows per question and the accuracy above which answering pays, a table that answers back shows which rules ask you to change how you work, and the last section says which official document to read for each test.

Six tests, six different shapes of paper

Each test publishes its pattern in its own document: CAT 2026 in its press release and FAQs, XAT in its instructions (the latest paper described there is XAT 2026), NMAT on GMAC’s pages, SNAP 2026, CMAT 2026 and MAT in their bulletins and FAQs. Put side by side, the six differ in every respect:

Each test in two sentences, with the rule that stands out:

• CAT 2026, run by the IIMs, gives each of VARC, DILR and QA exactly 40 minutes in a fixed order, with no switching while a section is open. It marks MCQs +3 and −1, and typed answers +3 and 0, and its documents give no question counts.
• XAT, run by XLRI, lasts three hours and, says its 2027 FAQ, has no sectional time limits. Its 2026 paper had 95 questions with five options each, and took 0.10 off for every Part 1 question left blank after the eighth.
• NMAT, run by GMAC, gives 108 questions in 120 minutes, in three sections whose order you choose, each with its own time limit. It is an adaptive testadaptive test: A test that chooses each next question according to how the candidate answered the earlier ones. More in “NMAT: section order, adaptivity and retakes” with no negative marking, reported as a scaled scorescaled score: Raw marks converted onto a fixed scale so that different sessions or test forms can be compared; each test publishes its own scale, such as NMAT's 36 to 360. More in “NMAT: section order, adaptivity and retakes” from 36 to 360, and you may sit it up to three times in a testing cycle.
• SNAP 2026, run by Symbiosis International University, has 60 questions, 60 marks and 60 minutes, with four responses per question and a quarter of a mark lost for a wrong one. You may take up to three tests, and the higher score counts.
• CMAT 2026, run by the National Testing Agency, has 100 questions for 400 marks in 180 minutes, five sections of 20. A right answer earns 4 and a wrong one costs 1, and the paper is in English only.
• MAT, run by AIMA, has 150 questions in 120 minutes, five sections of 30, on paper, on computer or both. Its FAQ states a 0.25 penalty for a wrong answer but not the mark for a right one, and MAT is held in February, May, September and December.

The program below holds the same facts in tests.py, each test with the document it comes from, and prints the size of every paper with its order of sections. The next two sections run two more programs on the same file:

"""Six published patterns side by side: the size of each paper and how its sections are ordered."""
from tests import TESTS

for t in TESTS:
    assert t["source"].startswith("https://"), f"no source for {t['test']}"

print("Test            Minutes  Questions  Sections  Order of sections")
for t in TESTS:
    counts = [q for _, q, _ in t["parts"]]
    questions = "not given" if None in counts else str(sum(counts))
    print(f"{t['test']:<16}{t['minutes']:<9}{questions:<11}{len(t['parts']):<10}{t['order']}")

Test            Minutes  Questions  Sections  Order of sections
CAT 2026        120      not given  3         fixed, the same for everyone
XAT 2026 paper  180      95         4         free within Part 1
NMAT            120      108        3         you choose it at the start
SNAP 2026       60       60         4         any section in any order
CMAT 2026       180      100        5         not stated in the bulletin
MAT             120      150        5         not stated in the FAQ

"""Six MBA entrance tests as their own documents describe them, each with the document it comes from.

None means the document does not say: CAT publishes no question counts, and MAT's FAQ gives the penalty for a wrong
answer but not the mark for a right one. XAT's paper is the latest one its instructions describe, XAT 2026.
"""

# A part: (name, questions, minutes or None when the part has no time of its own)
TESTS = [
    {
        "test": "CAT 2026",
        "by": "the IIMs",
        "source": "https://cdn.digialm.com/per/g06/pub/32842/EForms/image/CAT2026/CAT_2026_Press_Release_26-07-2026.pdf",
        "minutes": 120,
        "parts": [("VARC", None, 40), ("DILR", None, 40), ("QA", None, 40)],
        "marking": [("MCQ", 3, 1), ("typed answer", 3, 0)],
        "order": "fixed, the same for everyone",
    },
    {
        "test": "XAT 2026 paper",
        "by": "XLRI",
        "source": "https://xatonline.in/assets/pdf/Important_Instructions_XAT_2026.pdf",
        "minutes": 180,
        "parts": [
            ("Quantitative Aptitude & Data Interpretation", 28, None),
            ("Decision Making", 21, None),
            ("Verbal Ability & Logical Reasoning", 26, None),
            ("General Knowledge", 20, 10),
        ],
        "part_one_minutes": 170,
        "marking": [("Part 1", 1, 0.25), ("General Knowledge", 1, 0)],
        "skip_penalty": (8, 0.10),
        "order": "free within Part 1",
    },
    {
        "test": "NMAT",
        "by": "GMAC",
        "source": "https://www.mba.com/exams/nmat/about",
        "minutes": 120,
        "parts": [("Language Skills", 36, 28), ("Quantitative Skills", 36, 52), ("Logical Reasoning", 36, 40)],
        "marking": [("every question", None, 0)],
        "order": "you choose it at the start",
    },
    {
        "test": "SNAP 2026",
        "by": "Symbiosis International University",
        "source": "https://www.snaptest.org/documents/SNAP-2026-Bulletin.pdf",
        "minutes": 60,
        "parts": [
            ("General English", 10, None),
            ("Analytical & Logical Reasoning", 20, None),
            ("Quantitative, Data Interpretation & Data Sufficiency", 20, None),
            ("Ethics, Morality and Values", 10, None),
        ],
        "marking": [("every question", 1, 0.25)],
        "order": "any section in any order",
    },
    {
        "test": "CMAT 2026",
        "by": "the NTA",
        "source": "https://cdnbbsr.s3waas.gov.in/s3a381c2c35c9157f6b67fd07d5a200ae1/uploads/2025/10/202510171049586023.pdf",
        "minutes": 180,
        "parts": [
            ("Quantitative Techniques and Data Interpretation", 20, None),
            ("Logical Reasoning", 20, None),
            ("Language Comprehension", 20, None),
            ("General Awareness", 20, None),
            ("Innovation & Entrepreneurship", 20, None),
        ],
        "marking": [("every question", 4, 1)],
        "order": "not stated in the bulletin",
    },
    {
        "test": "MAT",
        "by": "AIMA",
        "source": "https://aima.in/mat/faq/",
        "minutes": 120,
        "parts": [
            ("Language Comprehension", 30, None),
            ("Intelligence & Critical Reasoning", 30, None),
            ("Mathematical Skills", 30, None),
            ("Data Analysis & Sufficiency", 30, None),
            ("Economic & Business Environment", 30, None),
        ],
        "marking": [("every question", None, 0.25)],
        "order": "not stated in the FAQ",
    },
]

Time per question: from 30 seconds to over two minutes

Dividing a paper’s time by its questions gives the time it allows for each one. MAT’s 120 minutes for 150 questions come to 7,200 ÷ 150 = 48 seconds; CMAT’s 180 minutes for 100 questions come to 108 seconds. Where a section has a clock of its own, it gets its own figure: NMAT’s Language Skills allow 46.7 seconds a question and its Quantitative Skills 86.7. The next program works out every figure from tests.py and sorts them from the fastest to the slowest:

"""The time each published pattern allows per question: a paper's minutes over its questions, or a part's own minutes
over its own questions where the part has a clock of its own. CAT publishes no counts, so it gets no figure."""
from fractions import Fraction

from tests import TESTS


def seconds(minutes, questions):
    return Fraction(minutes * 60, questions)


rows = []
for t in TESTS:
    counts = [q for _, q, _ in t["parts"]]
    if None in counts:
        continue
    if t["test"] == "XAT 2026 paper":
        rows.append(("XAT Part 1", seconds(t["part_one_minutes"], sum(counts[:3]))))
        rows.append(("XAT General Knowledge", seconds(t["parts"][3][2], counts[3])))
    elif all(minutes for _, _, minutes in t["parts"]):
        for name, q, minutes in t["parts"]:
            rows.append((f"NMAT {name}", seconds(minutes, q)))
    else:
        rows.append((t["test"], seconds(t["minutes"], sum(counts))))

print("Seconds per question, fastest first (one # for every 10 seconds)")
for name, s in sorted(rows, key=lambda r: r[1]):
    print(f"  {name:<26}{float(s):>6.1f} s  {'#' * round(s / 10)}")

Seconds per question, fastest first (one # for every 10 seconds)
  XAT General Knowledge       30.0 s  ###
  NMAT Language Skills        46.7 s  #####
  MAT                         48.0 s  #####
  SNAP 2026                   60.0 s  ######
  NMAT Logical Reasoning      66.7 s  #######
  NMAT Quantitative Skills    86.7 s  #########
  CMAT 2026                  108.0 s  ###########
  XAT Part 1                 136.0 s  ##############

Three figures stand out:

• XAT’s General Knowledge is the fastest, 30 seconds a question: 20 questions in 10 minutes.
• XAT’s Part 1 is the slowest, 136 seconds: 75 questions in 170 minutes, with decisions and data to work through.
• CAT has no figure, because its documents publish no question counts. A number you see for it is an estimate.

These figures are a planning tool, not a measure of difficulty. They say how fast the paper expects you to move, but not how hard its questions are, how a wrong answer is marked, or how many questions you should attempt.

The rules that change how you work

Beyond the content, four rules decide how you should practise for each test. The section clock decides whether a minute saved in one section can help another: in CAT and NMAT it cannot, because each section has its own clock. The marking decides when answering a doubtful question pays, and this program works out where, scheme by scheme:

"""When answering a doubtful question pays, test by test: the chance of being right above which an answer adds marks
on average, which is the penalty divided by the mark for a right answer plus the penalty."""
from fractions import Fraction

from tests import TESTS

print("Answering adds marks on average above this chance of being right")
for t in TESTS:
    for kind, right, wrong in t["marking"]:
        if wrong == 0:
            verdict = "no penalty: an answer can only add marks"
        elif right is None:
            verdict = "the mark for a right answer is not published"
        else:
            w = Fraction(str(wrong))
            verdict = f"{float(w / (right + w)):.0%} (lose {wrong}, gain {right})"
        print(f"  {t['test']:<16}{kind:<19}{verdict}")
blanks, cost = TESTS[1]["skip_penalty"]
print(f"  XAT also takes {cost:.2f} off for each question of Part 1 left blank after the first {blanks}.")

Answering adds marks on average above this chance of being right
  CAT 2026        MCQ                25% (lose 1, gain 3)
  CAT 2026        typed answer       no penalty: an answer can only add marks
  XAT 2026 paper  Part 1             20% (lose 0.25, gain 1)
  XAT 2026 paper  General Knowledge  no penalty: an answer can only add marks
  NMAT            every question     no penalty: an answer can only add marks
  SNAP 2026       every question     20% (lose 0.25, gain 1)
  CMAT 2026       every question     20% (lose 1, gain 4)
  MAT             every question     the mark for a right answer is not published
  XAT also takes 0.10 off for each question of Part 1 left blank after the first 8.

The break-even is 25 % for a CAT MCQ and 20 % for XAT, SNAP and CMAT, while in NMAT an answer can only add marks. Blanks cost nothing except in XAT. And the number of sittings decides how much one bad day costs: CAT, XAT and CMAT come once a year, while NMAT allows three attempts in a testing cycle, SNAP three tests and MAT four windows a year.

Answer the four questions in the table below from your own practice papers. It marks the tests whose rules ask you to change how you work, and says why with each test’s own numbers. No answer rules a test out: which tests you take depends on the institutes you apply to.

In short: plan inside each section for CAT and NMAT, answer doubtful questions only above each test’s break-even, watch your blanks in XAT, and sit NMAT or SNAP early enough to use a second attempt.

Which official document answers which question

A rule that changes from one year to the next is only as current as the document you read it in. Each test has one home for its rules:

Two cautions apply when you read them. First, documents of different years can disagree: XAT’s 2027 FAQ says there are no sectional time limits, while its 2026 instructions split the paper into two timed parts, 170 minutes and then 10 for General Knowledge, so this lesson quotes each with its year until the 2027 instructions appear. Second, a document can say two things at once: SNAP’s 2026 Bulletin says there is no normalization of scores, and also that the papers will be normalised for difficulty across the three tests, so read both lines rather than guess a method.

Practise: six tests side by side

Eleven questions cover the patterns, the time per question, the rules and the documents. Your weakest part comes first, a right answer brings a harder question, and a question you miss returns before the end:

Interview questions

Warm-up: which of the six tests lets you choose the order of its sections?

NMAT does it most explicitly: GMAC says candidates choose the order of the three sections at the start of the exam, and each section then runs on its own clock. SNAP 2026’s instructions say a candidate may attempt any section in any order, within its single 60-minute clock. XAT’s 2027 FAQ says every section is accessible throughout the exam, with no sectional time limits; its 2026 paper allowed free movement among the three sections of Part 1, with General Knowledge after them. CAT is the opposite: the order is fixed and the same for every candidate, VARC, then DILR, then QA, each with exactly 40 minutes. The CMAT 2026 bulletin and the MAT FAQ state no rule about the order of sections.

Key takeaways

• The six papers differ in every respect: 60 minutes for SNAP and 180 for XAT and CMAT; 60 questions in SNAP and 150 in MAT; and CAT publishes no count at all.
• Time per question runs from 30 seconds in XAT’s General Knowledge to 136 in XAT’s Part 1, with MAT at 48, SNAP at 60 and CMAT at 108, but the figure says nothing about difficulty or marking.
• A wrong answer costs a third of a right answer on a CAT MCQ, a quarter in XAT, SNAP and CMAT, and nothing in NMAT, so the break-even accuracy is 25 % for a CAT MCQ and 20 % for the other three.
• CAT and NMAT give each section its own clock, XAT penalises blanks after the eighth, and NMAT, SNAP and MAT offer more than one sitting a season while CAT, XAT and CMAT do not.
• Each test’s rules live in its own documents, and documents of different years, or lines of the same document, can disagree: quote each with its year.