Roadmap

What is planned for the Evaluation API but not available yet, and what to do in the meantime.

Everything on this page is planned, not available. There are no dates, and plans can change. Build only on what the API reference and the Changelog list as live today. If one of these matters for your launch, tell us at [e-mail]: partner demand sets the order.

Planned

Webhooks

Signed event notifications when a submission is graded, fails or is finalized, so you can stop polling.

Today: poll GET /submissions?updated_since= as shown in Keeping in sync.

Bulk batches with name matching

Upload one scan of a whole bundle of answer sheets and let Evalezy split it and match each copy to a candidate by name or roll number.

Today: upload one PDF per answer sheet and create one submission per candidate.

Phone photos

Submit answer sheets as photos (images[]) instead of a scanned PDF.

Today: convert photos into one PDF per answer sheet. Sending images answers 422 feature_not_available.

Exam from a question-paper PDF

Create a draft exam by uploading the question paper and letting Evalezy read the questions and marks.

Today: send questions as JSON in POST /exams.

Rubric generation on request and locking

Generate a draft rubric for each question from the model answer, then lock rubrics before grading starts.

Today: send your own rubric per question, or let the AI grade against the model answer. If a long answer has neither, a rubric is generated from the question text on the first copy and reused (source: "generated"); review it with GET /exams/{id}/rubrics.

Hosted review links

A secure link your teachers can open to review and adjust AI marks.

Today: build review into your product with the review endpoints, or have teachers review in the Vacademy dashboard, where API exams appear tagged Source: API.

CSV results

GET /exams/{id}/results?format=csv for a spreadsheet of an exam’s marks.

Today: page through the JSON results (50 per call) and build the CSV on your side. format=csv answers 422 feature_not_available.

Hindi and regional languages

Grading answers written in Hindi and other Indian languages, typed or handwritten.

Today: English only. A typed answer (long answer or one-word) in which more than 20% of the letters are Devanagari is refused with 422 language_not_supported; handwritten Hindi copies fail with error.code: "language_not_supported" and are not charged.

SDKs

Official client libraries.

Today: call the REST API with any HTTP client. The OpenAPI document at https://api.evalezy.com/v1/openapi.json can generate a client in most languages.

In beta

Choice groups ("attempt any N of M")

Papers with internal choice, where a candidate answers any N of the M questions in a group and only N count towards the total. The endpoint exists but is switched off by default and answers 422 feature_not_available. It is enabled on request for pilot partners: write to [e-mail].

Today: send papers without internal choice.

Also not available yet

These parts of the request shape are reserved and answer 422 feature_not_available:

Request	Today
files[] on POST /exams/{id}/submissions (answer booklet plus supplementary sheets)	Merge the booklet and supplements into one PDF and send one upload_id.
question_ids on POST /submissions/{id}/re-evaluate (re-grade single questions)	Re-evaluate the whole copy, or override the questions by hand with PATCH /submissions/{id}/questions/{question_id}.
purge=true on DELETE /exams/{id}	Delete without purge.

See Errors for how feature_not_available is returned, and the Changelog for what has shipped.