How many seats, how many students

1. Executive summary

EduPeak's home page shows twelve doors. This is the research behind them.

The twelve are the admission routes realistically open to a Class 11 or 12 student carrying Physics, Chemistry and Mathematics: state engineering, the national engineering system, the research institutes, architecture, the merchant navy, the defence academies, the university system through CUET, and the ordinary BSc in the student's own district. For each one we asked the same two questions. How many seats? How many people applied?

Three findings shape everything else.

The range is enormous, and it is a range across doors, not across students. At one end, a BSc seat in a state or autonomous college is close to open admission on board marks — roughly one candidate per seat, with seats left over. At the other, the National Defence Academy runs at something like 1,300 applications per vacancy. That is a factor of more than a thousand between two doors that the same student may walk through in the same week.

Most of the seats are not where most of the attention is. The routes that carry the fewest seats attract the most public conversation, coaching spend and family anxiety. The routes carrying hundreds of thousands of seats — state engineering, CUET, the plain BSc — are discussed least. Thousands of state engineering seats go vacant every year while the same families describe the system as impossibly full.

The numbers moved sharply between 2025 and 2026, in one direction. JEE Main registrations reached 16,04,854. MHT-CET's PCM group grew from about 4.64 lakh to 7,32,268 in a single year. IAT registrations roughly doubled. Seats grew too — JoSAA added 4,470 — but not at that pace. A ratio quoted from a 2024 article is not describing 2026.

This version corrects six figures in the published note, including the IIT seat count and the NDA vacancy number, and re-states two ratios that a year of growth had moved a long way. Section 10 lists every correction.

A candidates-per-seat ratio is a property of a door, never of a student. It describes what happened at a counter last year. It says nothing about the person reading it.

2. At a glance

The twelve routes, in the order the doors animation uses — least selective first. "Seats" and "candidates" describe the most recent cycle for which the number is public, named in the row. Estimates are labelled in the row itself.

3. How to read a ratio

This section matters more than any single number in the paper, so it comes before the detail rather than after it.

What a ratio is. Candidates who applied, divided by seats that existed, for one route in one completed cycle. Both numbers are counts of things that already happened.

What a ratio is not.

It is not a probability. "33 candidates per seat" does not mean any individual has a 1-in-33 chance. Candidates are not identical and seats are not allocated by lottery. The ratio is an aggregate; a chance is a personal quantity, and we do not compute personal quantities.

It is not a difficulty rating. A route can have a high ratio because it has very few seats rather than because its examination is hard. Route 12 is the clearest case: 394 vacancies is a small number, and the ratio is mostly a statement about that.

It is not comparable across routes without care. Some counts are registrations, some are candidates who actually appeared, and the gap is not small — 16,04,854 registered for JEE Main 2026 and 15,38,468 appeared. Some seat counts are one institution's published matrix; some are a national estimate nobody officially maintains. A ratio built from a registration count and an estimated seat count is a rougher object than one built from two official counts, even though both are printed as a single number.

It is not stable between years. Section 5 shows how far some of these moved in twelve months.

And it is not about you. A student reads "85 per seat" and hears "you are one of 85". That is not what the sentence says. It says a counter had one seat and 85 forms arrived at it. The student is not in the number. Nothing on this page, and nothing that could be built on it, licenses the step from a ratio to a person — or from a score to a destination. We do not take that step here and we will not take it elsewhere.

4. The twelve routes in detail

4.1 BSc in your own state · Class 12 marks, no entrance exam · ~1

Open admission on board marks at most state and autonomous colleges. There is no entrance examination, no coaching requirement and no ratio worth quoting, because in most places the seats are not the binding constraint. AISHE has recorded roughly 49 lakh students enrolled in BSc programmes nationally. This is by a wide margin the largest door in the set and the one least written about.

For a student in a district without an engineering college within reach, this is not a fallback. It is a first-class route into a science degree, and it is the route most of India actually takes.

4.2 State engineering · MHT-CET, KCET, WBJEE and others · ~2–5

The largest engineering door by seat count, and the one where vacancy is a real and recurring feature.

Maharashtra: MHT-CET 2026 drew 7,32,268 registrations in the PCM group alone, against roughly 1.8 lakh engineering seats. That PCM figure was about 4.64 lakh in 2025 — a rise of more than half in one year, and the sharpest single movement in this paper. Karnataka: KEA's 2026 seat matrix runs to over 86,000 seats, of which 66,000+ are engineering. West Bengal: 1,01,643 candidates appeared for WBJEE 2025 against roughly 52,000 seats; WBJEEB had not published 2026 statistics at the research date.

Thousands of seats in this group go vacant every year, in every one of these states, after every counselling round. Both facts are true at once: the ratio has risen, and seats remain unfilled at the end.

4.3 Architecture · NATA, JEE Main Paper 2A · ~3–6

About 63,000 registered for JEE Main Paper 2A in 2025, against roughly 20,000 to 24,000 sanctioned B.Arch seats nationally. Actual admissions run near 11,000, so a large share of sanctioned seats stay unfilled year after year.

This is the clearest example in the paper of a door where the ratio and the experience diverge. On paper it is competitive. In practice, seats are going empty.

4.4 250+ universities · CUET-UG · ~5

CUET-UG 2026 drew 15,68,866 registrations. Against an estimated 3 lakh undergraduate seats across participating universities, that is about five candidates per seat — but the seat figure is the weakest number in this paper. No authority publishes a consolidated CUET seat total. Each participating university publishes its own intake, and the count of participating universities itself is reported variously between 230 and 250.

Treat this row as an order of magnitude and nothing finer.

4.5 Merchant navy · IMU-CET · varies

Reported applicant counts range widely — from 25,000 to about 1 lakh, depending on the source and on whether affiliated-institute applications are counted. Indian Maritime University's own campuses hold roughly 966 undergraduate seats; affiliated institutes add several thousand more.

No ratio is quoted because the applicant number is not reliable enough to divide by anything. Medical and eyesight standards apply here and do not apply to any other route in this paper, which makes the seat count only part of the eligibility picture.

4.6 NITs, IIITs and GFTIs · JEE Main · ~33

JEE Main 2026: 16,04,854 unique candidates registered across both sessions; 15,38,468 appeared. Non-IIT seats allocated through JoSAA 2026: 48,372 — NITs 25,162, IIITs 11,518, GFTIs 11,692. On registrations that is about 33 per seat; on appearances, about 32.

JoSAA 2026 was a growth year. Total seats rose from 62,853 to 67,323, an increase of 4,470, and the number of participating GFTIs went from 47 to 56.

4.7 IISERs, NISER and CEBS · IAT, NEST · ~85

This row moved further than any other since version 1.0, and the movement is worth stating plainly.

IAT 2026 drew over 2 lakh registrations, with about 1.86 lakh appearing. In 2025 the figures were 1,49,228 registered and 1,19,196 appeared; in 2024, 98,000 registered. Registrations have roughly doubled in a year and roughly quadrupled in three. Seats through IAT total 2,343 — 2,123 BS-MS places across the seven IISERs plus around 220 B.Tech and BS places. That is about 85 registered candidates per seat, against the roughly 42 this note published in August 2026 on the older figures.

NEST covers 257 regular seats — NISER Bhubaneswar 200 and UM-DAE CEBS Mumbai 57, rising to 261 with supernumerary places. NEST 2026 registration statistics had not been published at the research date; the most recent published figure is over 45,000 for 2024.

IAT registrations roughly doubled between 2025 and 2026 while seats stayed flat. Any ratio for this route quoted from a 2024 or 2025 source is now about half of what it should be.

4.8 IITs · JEE Advanced, after JEE Main · ~85

Two ratios sit on this route and they measure different things.

Across the whole funnel: 16,04,854 JEE Main 2026 registrations against 18,826 IIT seats through JoSAA 2026 — about 85 per seat.

Within JEE Advanced itself: 1,87,389 registered for JEE Advanced 2026, of whom 1,79,694 appeared in both papers and 56,880 qualified. Against 18,826 seats that is about 10 per seat at that stage.

The second number is not a smaller version of the first. It describes a different population — candidates who had already cleared the JEE Main cutoff to be eligible. Quoting one where the other belongs is the most common arithmetic error made about this route.

4.9 BITS Pilani · BITSAT · ~70–100

Around 2 to 3 lakh applications by public estimate, against roughly 2,970 first-degree seats across Pilani, Goa and Hyderabad. BITS does not publish application counts. The ratio is therefore a range built on an estimate divided by an estimate, and it is presented as a range for that reason.

4.10 ISI and CMI · own entrance, Mathematics only · not published

Roughly 190 to 210 seats between them. ISI offers 79 places in B.Stat at Kolkata and 79 in B.Math at Bengaluru — 63 regular plus 16 supernumerary for female candidates in each, with up to 5 further INMO-channel places. CMI's intake is smaller, roughly 30 to 50.

Neither institution publishes applicant counts, so no ratio is quoted. This row exists to say that. A door with no published denominator is a real door; it is simply one about which this particular arithmetic cannot be done, and filling the gap with a guess would make the whole table less trustworthy rather than more complete.

Where no applicant count is published, we print no ratio. A guessed denominator would make every other row in the table less believable.

4.11 IISc and IIST · JEE Advanced, IAT · several hundred

IISc Bengaluru's four-year BS (Research) programme has 125 places in 2026 — 100 for Indian nationals and OCI (I), 20 supernumerary for female candidates, 5 for foreign nationals — admitting on JEE Advanced 2026 or IAT 2026 rank. IISc entered JoSAA for the first time in 2026, which is where those 125 seats appear in the seat matrix.

IIST Thiruvananthapuram admits solely on JEE Advanced rank, with about 140 regular places: 60 in Aerospace Engineering, 60 in Electronics and Communication / Avionics, 20 in the dual degree, plus supernumerary and PMSSS allocations.

Together that is around 265 seats drawn from the 1,79,694 candidates who appeared in JEE Advanced 2026 — of the order of several hundred candidates per seat. Version 1.0 gave ~400 on a seat count of ~450; both the seat figure and the ratio are revised here.

4.12 NDA and Naval Academy · UPSC NDA exam · ~1,300

394 vacancies per examination in 2026 — the same figure for NDA 1 and NDA 2 — and the examination runs twice a year. NDA 2 2025 carried 400. Version 1.0 quoted 406, which appears to be an earlier 2025 cycle.

Against an estimated 5 to 6 lakh applications per cycle, that is roughly 1,300 per vacancy, the highest ratio in this paper. Vacancies are official UPSC numbers. Application counts are estimates — UPSC does not routinely publish NDA application totals.

This route also carries medical, physical and age standards that no other route here carries, and a Service Selection Board stage after the written examination. The seat arithmetic is only one part of the gate.

5. What the pattern shows

The spread is the finding. From roughly 1 candidate per seat to roughly 1,300, across twelve doors that are simultaneously open to the same student. No other fact in this paper is as large.

Seats and attention are inversely arranged. The four routes with the most seats — BSc, state engineering, CUET, architecture — occupy the smallest share of public conversation. The three with the fewest — IISc and IIST, ISI and CMI, NDA — occupy a large share of it. A student who forms their picture of Indian admissions from what is discussed around them will systematically overestimate how full the system is.

Vacancy and competition coexist. State engineering seats go unfilled every year while candidate numbers rise. B.Arch admits about 11,000 against 20,000-plus sanctioned seats. Both things are simply true, and the reason is distribution — the vacancies and the crowd are rarely at the same college.

The candidate side moved faster than the seat side in 2026. JEE Main reached its highest recorded registration. MHT-CET PCM rose 58 per cent in a year. IAT roughly doubled. CUET-UG passed 15.5 lakh. JoSAA added 4,470 seats and IISERs added none. This is why the paper is dated to a cycle rather than presented as a standing fact.

A number ages faster than it looks. Two ratios in the published note — routes 7 and 11 — had drifted by roughly a factor of two within a year of the underlying data. Neither was wrong when written.

The routes with the most seats get the least attention, and the routes with the fewest get the most. A student who learns the system from the conversation around them will get it backwards.

6. What this means for preparation

Read this section as a description of what the numbers imply for how a PCM foundation is built. It is not a recommendation about where to apply, and it deliberately contains no advice of that kind.

One PCM foundation feeds all twelve of these doors. Physics, Chemistry and Mathematics at Class 11 and 12 level is the shared input to routes 2, 3, 6, 8, 9 and 11; to routes 1 and 4 through their science subjects; to route 7 through PCM alongside Biology; to routes 5 and 12 through Mathematics and school science alongside the other sections those papers carry; and to route 10 through Mathematics. The proportions differ. The subject does not. This is one body of work with different edges on it, not twelve separate preparations — and the chapters a student is already obliged to do for the board are the same chapters every one of these routes is built on.

The same Class 11–12 PCM sits behind every door in this table. A student is not choosing between twelve preparations; they are doing one, and the differences between the routes are in pattern, pace, scope and marking at the edges — which is what the seven companion papers set out, route by route.

Board marks are load-bearing on more routes than students expect. Route 1 runs entirely on them. Several state routes in route 2 put them directly into the merit rank — KCET at half, AP EAPCET at a quarter, and Gujarat at a weighted share of the ACPC rank that has itself been published differently in different years. Route 9 uses them as a hard eligibility gate. The seven exam papers in section 8 set out exactly how each one counts them.

The rules differ more than the syllabus does. Negative marking, whether both school years are examinable, how ties are broken, whether board marks enter the rank — these vary sharply between routes that examine nearly the same content. That is where a student loses marks they have already earned.

A published ratio is a fact about last year. Re-check the ones that matter to you against the current bulletin before the cycle you are sitting.

Nothing here should narrow your options this early. Twelve doors are open. This paper counts them; it does not close any of them, and it is not equipped to.

The spread cuts the encouraging way as well as the discouraging one. Some of these doors are very narrow — 394 vacancies is 394 vacancies, and no amount of framing changes it. But the narrowness belongs to those doors, not to the twelve. The routes carrying hundreds of thousands of seats are on the same table, built on the same subjects, and several of them have seats left over at the end of counselling. The effort a student puts into a chapter of Physics is not spent on one narrow route and lost if that route does not open. It is the same chapter every one of the twelve is asking about.

7. Common misconceptions

8. How this paper relates to the seven exam papers

This is the overview. Seven companion papers are the detail, and each one takes a single route from the table above and works through it properly — the pattern, the marking scheme, the syllabus gaps against JEE Main, how board marks enter the rank, and what a student preparing for one while preparing for another actually has to change.

- MHT-CET PCM and JEE Main PCM — Maharashtra. Route 2.

- KCET PCM and JEE Main PCM — Karnataka. Route 2.

- COMEDK UGET and JEE Main PCM — Karnataka private institutes. Route 2.

- WBJEE PCM and JEE Main PCM — West Bengal. Route 2.

- GUJCET PCM and JEE Main PCM — Gujarat. Route 2.

- AP EAPCET, TG EAPCET and JEE Main PCM — Andhra Pradesh and Telangana. Route 2.

- BITSAT PCM and JEE Main PCM — national. Route 9.

Each of those seven carries the same nine-row table showing how board marks count across the engineering entrance routes, so a reader who opens two of them meets the same frame twice. This paper carries the counting; they carry the rules.

Read together, the eight documents make one argument. The landscape is wider than it is usually described, the doors differ from one another in rules far more than in content, and none of that has anything to do with which student is reading.

9. How EduPeak's PCM pattern fits

EduPeak's core work is a Grade 11 and 12 PCM learning framework: chapters decomposed into key concepts, learning outcomes and assessment indicators, each indicator tagged to a cognitive level, with chapter-level diagnostics reported in plain language as Strong, Developing or Needs work.

That framework is upstream of every door in this table, which is the reason this paper exists in the shape it does. A chapter is a chapter whichever counter the student eventually walks up to. The framework measures what a student has learned; it does not rank students against a route, and it is not built to.

What the framework already supports. Both school years, decomposed and testable. Chapter-level diagnostics that speak to route 1 and to the state routes where board marks enter the rank directly. Cognitive-level tagging, so the same bank can be re-weighted toward a route's centre of gravity without content being re-authored. Pattern-honest assessment, which reproduces real marking rather than flattering the student.

What this paper is deliberately not used for. The framework does not map a diagnostic to a door. It will not say a student is "IIT material" or "better suited to a state college". There is no path from a chapter-level readiness signal to a destination that would be honest, and building one would make the diagnostic worse — a student told early which door is theirs stops preparing for the other eleven.

Where this leads. The table above is a landscape a family can read once and understand. Making it available in Indian languages, and keeping it current cycle by cycle, is worth more than adding a thirteenth column to it. A student in a district town should be able to see the same twelve doors, with the same numbers, as a student anywhere else. That is the whole design intent of this paper.

EduPeak's diagnostics measure what a student has learned. They do not, and will not, map that to a door. There is no honest path from a chapter-level readiness signal to a destination.

What a student reading this table already has. This paper has spent ten sections on how many people stand in front of each door, so it should end on the other side of the arithmetic, which is just as factual and gets said far less often.

The twelve doors run on one subject. A student carrying Class 11 and 12 Physics, Chemistry and Mathematics is doing the work all twelve are built on, in the two years they were going to spend on it regardless. That work is compulsory: the board examination requires it whatever else is or is not attempted. So the question is not whether to do it. It is whether a student can see clearly where they stand in it — which chapters are solid, which are not, and which one to open tomorrow.

That is the whole of what EduPeak's chapter and concept decomposition, Bloom-tagged assessment indicators and chapter-level diagnostics are for. They are built on the same Class 11–12 PCM syllabus the school teaches, which means a large portion of the mandatory PCM preparation is already covered by one honest, chapter-level view of where a student stands. EduPeak does not teach it — school and coaching class teach. This measures, and shows what to work on next.

None of that is a promise about an outcome, and this paper is the last place that would make one. It is a structural fact about the landscape, and it points one way: the ground is more familiar than the table makes it look, the effort compounds across every row of it, and a student who works steadily at one body of PCM is not making a bet on a single narrow door. They are doing the thing all twelve are asking for.

Want to know where you stand — not against these ratios, but against the syllabus itself? That is what EduPeak measures: chapter by chapter, concept by concept, from a readiness check before Grade 11 to a growing library of full-length mocks at the end. Create your account and take Basecamp — it is the right first step whichever of these twelve doors you eventually choose.

10. Document information

Version 2.0. Researched 20 August 2026; republished 22 August 2026.

Created by EduPeak (with help from AI). Subject to correction.

Cycle. Figures describe the 2024, 2025 and 2026 admission cycles as indicated in each row. The 2026 cycle has now closed — JEE Main results were declared in April, JEE Advanced on 1 June, JoSAA allocation and state counselling ran through July and August. Every reader is a 2027 or later candidate. The 2026 position is the best available guide to 2027 and is not a substitute for the 2027 notifications.

Corrections in this version.

1. IIT seats. Version 1.0 gave 18,951 seats through JoSAA 2026. The correct IIT figure is 18,826; the extra 125 are IISc's, which entered the JoSAA matrix for the first time in 2026.

2. NDA vacancies. Version 1.0 gave 406 per examination in 2025. Both NDA 1 and NDA 2 in 2026 carry 394; NDA 2 2025 carried 400.

3. Route 7 ratio. Version 1.0 gave ~42 on IAT 2025 figures described as "over 1 lakh registered for 2,363 seats". IAT 2025 actually drew 1,49,228 registrations, and IAT 2026 drew over 2 lakh against 2,343 seats. The ratio is now about 85.

4. Route 11 seats and ratio. Version 1.0 gave ~450 seats (IISc ~250, IIST ~200) and ~400 per seat. IISc's BS (Research) intake is 125; IIST's regular intake is about 140. The ratio is materially higher than stated and is now given as an order of magnitude rather than a single number.

5. ISI seats. Version 1.0 gave 168. The published matrix is 79 in B.Stat and 79 in B.Math, totalling 158 including supernumerary places, plus up to 5 INMO-channel places per programme.

6. Route 2 candidate figures. MHT-CET's PCM group has grown from about 4.64 lakh in 2025 to 7,32,268 in 2026. KCET's 2026 seat matrix gives 66,000+ engineering seats, against the 77,000+ quoted in version 1.0.

What could not be confirmed. The B.Arch registration and sanctioned-seat figures; IMU-CET applicant counts and the 966-seat figure; the CMI intake; NDA application totals; NEST 2026 registrations; WBJEE 2026 statistics; the consolidated CUET seat total and participating-university count; MHT-CET's and WBJEE's total seat figures. Each of these rows carries a @VERIFY marker naming the source that would settle it. None was removed, because a route with no public denominator is still a route a student can take.

Confirmed against primary sources. JEE Advanced 2026 registration, appearance and qualification counts, from the official results press release at jeeadv.ac.in. All other confirmations in this pass came from secondary reporting of official releases and are labelled as such below.

Standing rule. Selectivity in this paper attaches to the door, never to the student. This document describes the landscape identically for every reader. It does not suggest where anyone should aim and does not connect any score to any destination. If a future version of this paper appears to do either, that version is wrong.

11. Sources

Official — check these first.

1. Joint Seat Allocation Authority — seat matrix and business rules. https://josaa.nic.in

2. National Testing Agency — JEE (Main) bulletin, result statistics. https://jeemain.nta.nic.in

3. JEE (Advanced) 2026 — results press release, 1 June 2026. https://jeeadv.ac.in/documents/Result2026PressRelease.pdf

4. National Testing Agency — CUET-UG. https://cuet.nta.nic.in

5. Union Public Service Commission — NDA & NA examination notices. https://upsc.gov.in

6. All India Survey on Higher Education. https://aishe.gov.in

7. IISER Admissions — IAT information brochure and seat matrix. https://www.iiseradmission.in

8. NEST — NISER and UM-DAE CEBS. https://www.nestexam.in

9. Indian Institute of Science — BS (Research) admissions. https://www.iisc.ac.in/admissions

10. Indian Institute of Space Science and Technology. https://www.iist.ac.in

11. Indian Statistical Institute — admission business rules. https://admission.isical.ac.in

12. Chennai Mathematical Institute — admissions. https://www.cmi.ac.in/admissions

13. BITS Pilani — BITSAT admissions. https://www.bitsadmission.com

14. Indian Maritime University. https://www.imu.edu.in

15. Council of Architecture. https://www.coa.gov.in

16. State CET authorities — Maharashtra https://cetcell.mahacet.org · Karnataka https://cetonline.karnataka.gov.in/kea · West Bengal https://wbjeeb.nic.in

Secondary — used where an official page was not reachable during this pass, and labelled as such.

17. Careers360 — JoSAA 2026 seat matrix breakdown, 4,470 seats added. https://engineering.careers360.com/articles/josaa-2026-seat-matrix-4470-new-seats-added-iits-nits-iiits-gftis

18. Careers360 — IAT 2026 participation figures and year-on-year comparison. https://engineering.careers360.com/articles/how-many-students-appeared-for-iat-2026-iiser-participation-numbers-out

19. Careers360 — MHT-CET 2026 registration statistics. https://news.careers360.com/mht-cet-2026-registration-highest-ever-12-5-lakh-shows-36-percent-rise-in-one-year-pcm-pcb-mba-mms-exam-dates-cetcell-mahacet-org

20. Careers360 — IISc BS (Research) 2026 admissions and intake. https://news.careers360.com/iisc-bangalore-opens-registration-for-four-yea-bachelor-of-science-bs-research-admissions-2026-jee-advanced-iat-sat-act

21. Physics Wallah — NTA JEE Main 2026 session-wise registration and appearance data. https://www.pw.live/news/nta-jee-main-2026-data-out-how-many-students-appeared

22. Shiksha — JoSAA 2026 total of 67,323 seats. https://www.shiksha.com/news/engineering-josaa-2026-seat-matrix-released-total-seats-increase-to-67-323-across-iits-nits-iiits-and-gftis-blogId-231805

23. KollegeApply — CUET-UG 2026 registration total. https://news.kollegeapply.com/news/cuet-ug-registration-statistics-2026

24. EduPeak — companion white papers on MHT-CET, KCET, COMEDK, WBJEE, GUJCET, AP/TG EAPCET and BITSAT, August 2026.

A note on sources. Secondary sources here are reporting official releases, not originating figures, and every one of them is paired with a @VERIFY marker pointing at the official document that would settle the question. Where a secondary source and an official document differ, the official document is right, and this paper is wrong until corrected. If you are checking a number from this paper, open the official page — not a coaching website, and not this paper.