Industry2026-06-06·8 min read

CBSE OSM 2026: The February Warnings That Could Have Prevented a National Crisis

On 26 February 2026, a mandatory mock evaluation session exposed critical vulnerabilities in CBSE's on-screen marking system. Here is what those warnings contained, why they were ignored, and what digital evaluation teams must learn.

A Crisis That Announced Itself Weeks in Advance

By early June 2026, CBSE's Class 12 On-Screen Marking experiment had unravelled into what many observers are calling one of the most consequential examination failures in post-independence India. The national pass percentage dropped to 85.20 percent, the lowest in seven years, down sharply from 88.39 percent in 2025. Over 17 lakh students received results they could not trust. Four lakh-plus students requested scanned copies of their answer sheets. More than 40,000 submitted re-evaluation forms before the portal closed at midnight on June 6, 2026.

What makes this crisis particularly damaging for institutional trust is not the failure itself but the fact that the failure was visible weeks before results day. On 26 February 2026, a mandatory mock evaluation session conducted by CBSE for affiliated schools revealed precisely the vulnerabilities that later caused the system to collapse at scale. Teachers reported portal access failures. System performance was described as slow and unstable. Internet connectivity at school evaluation centres was demonstrably inadequate. Errors appeared in teacher data on the OASIS portal used to authenticate evaluators.

None of these warnings produced a course correction.

What the Mock Evaluation Revealed

The purpose of a pre-deployment mock evaluation is to stress-test the system under conditions close to real-world use before millions of answer books are committed to it. Industry practice in large-scale examination technology deployment calls for a mock that exercises the full end-to-end flow: login, answer sheet retrieval, mark entry, submission, and portal audit. When a mock reveals systemic failures — not isolated errors but failures affecting portal access, data integrity, and evaluator authentication simultaneously — the appropriate institutional response is a rollback decision or, at minimum, a delayed deployment with a documented remediation plan.

The February 26 mock session flagged:

Portal inaccessibility for significant numbers of evaluators during the session window

Slow system response times under concurrent user load

Inadequate internet infrastructure at school-based evaluation centres, which CBSE had designated as evaluation hubs to reduce travel burden on teachers

Inaccurate teacher data on the OASIS portal, meaning some evaluators could not authenticate themselves to access the system at all

These are not peripheral edge cases. Each of these failures points to a different layer of the technology stack: application performance, network dependency, data quality. A system that exhibits all four categories of failure simultaneously in a controlled mock environment — with a fraction of the eventual user load — has not met the minimum bar for deployment at scale.

The Deployment That Went Ahead Anyway

CBSE's maiden large-scale OSM deployment covered 98,66,622 answer books. Approximately 40 crore scanned pages were digitised and uploaded to the evaluation platform operated by Coempt Edu Teck (formerly Globarena Technologies), which had won the CBSE contract as the lowest qualified bidder.

The vendor's track record was a matter of public record before the contract was awarded. In 2019, Globarena Technologies had deployed examination software for Telangana state boards that caused a bubbling error in OMR sheets, resulting in 3.28 lakh students being failed incorrectly. That crisis was linked in subsequent reporting to 19 to 21 student suicides. In 2023, the same firm was linked to examination anomalies in Telangana's intermediate exams, associated with 19 additional student suicides in a single week. The company rebranded as Coempt Edu Teck, with its CEO describing the name change as a routine branding exercise.

After results were declared in May 2026, the cascade of failures matched precisely what the February mock had foreshadowed:

Students received answer sheets belonging to different students entirely

Blurred, illegible scans were uploaded as valid evaluation copies

Pages were missing from digitised answer book records

The re-evaluation portal initially quoted a fee of Rs 69,000 per subject before the error was corrected

The portal rejected valid payment submissions under load

IIT Madras and IIT Kanpur were called in to stabilise the platform

CBSE chairman Rahul Singh and secretary Himanshu Gupta were transferred to other departments. The Education Minister accepted responsibility for discrepancies while defending OSM as a progressive, student-centric initiative.

Five Pre-Deployment Checks That Should Be Non-Negotiable

The CBSE experience demonstrates that digital evaluation failures at scale are rarely sudden. They are the cumulative result of governance shortcuts applied upstream. The following checks, had they been enforced, would have produced a different outcome.

Minimum viable mock scale. A mock evaluation covering fewer than 10,000 scripts across at least 15 to 20 geographically distributed centres cannot generate statistically meaningful load data. The mock must simulate concurrent evaluator logins at the same density expected on day one of actual deployment.

Independent technical audit before go-live. Vendor self-certification is not sufficient for systems handling the educational records of millions of students. An independent technical team — not the procuring institution, not the vendor — should certify infrastructure capacity, failover mechanisms, and data integrity controls before the system is declared ready for production.

Parallel manual backup for the first deployment cycle. In a first-year deployment, physical answer sheets should be retained in secure storage until digital results are verified and the re-evaluation window has closed. This eliminates the scenario where a student's recourse is blocked by a digital failure with no physical backup available.

Evaluator connectivity pre-certification. If evaluation is conducted at school or college-based centres rather than centralised evaluation camps with dedicated infrastructure, each centre's connectivity must be tested and certified under simulated load conditions, not assumed to be adequate.

Phased rollout with a defined exit threshold. A first-year OSM deployment should cover a maximum of 20 to 30 percent of answer books, chosen across subject areas and regions, with a clear metric that would trigger full deployment in year two. Deploying 98,66,622 books in year one of a new system removes the capacity to course-correct.

The Ten-Point Reform Blueprint and What It Means

The crisis has generated a detailed 10-point reform proposal now circulating among examination governance bodies and attracting attention in parliamentary discussions. The core elements include a mandatory two-year phased pilot before any full-scale deployment, the establishment of an independent Examination Technology Audit Authority with powers to reject vendor go-live approvals, and the creation of a Parliamentary oversight committee for examination bodies.

Other proposals address the specific failure modes observed: reinstating post-result verification safety nets that were eliminated simultaneously with the new technology, establishing a separate Examination Grievance Redressal Tribunal with statutory timelines, and activating the Public Examinations (Prevention of Unfair Means) Act, 2024 as an enforcement tool.

The Supreme Court, reviewing petitions related to the crisis, observed that the National Testing Agency had appeared not to have learned from 2024 failures and that institutional trust in government submissions had been "betrayed" by 2026 repeat failures.

The Institutional Lesson

For universities and examination boards considering digital evaluation adoption, the CBSE experience is not an argument against the technology. It is an argument for the discipline required to deploy it responsibly. OSM works. It has worked at scale for years in jurisdictions as demanding as the UK's A-level system and across multiple Indian state boards that piloted the technology carefully before scaling.

What the February mock evaluation tried to communicate — and what the system failed to hear — is that technology deployment readiness is a technical determination, not a political or administrative one. No procurement timeline, no ministerial announcement, and no vendor assurance can substitute for a system that has actually demonstrated performance under realistic load.

The 85.20 percent pass rate will recover. What takes longer to rebuild is the confidence of 17 lakh students, their families, and the institutions that depend on examination credibility to function.