The cvpr 2027 submission deadline is November 16, 2026, at 11:59 PM Anywhere on Earth, and the call for papers says the quiet part out loud: the date is fixed, and no extension will be given. Paper enrollment closes six days earlier, on November 10. Between those two dates sits everything else this call changes, including a reviewer duty that now applies to every author and an AI policy that, by the call's own admission, is not finished yet.
The dates, exactly as the call prints them
The CVPR 2027 call for papers lists six dates, and every one of them is worth putting on a calendar today:
- Paper enrollment deadline: November 10, 2026, 11:59 PM AoE
- Submission deadline: November 16, 2026, 11:59 PM AoE
- Supplementary materials deadline: November 23, 2026, 11:59 PM AoE
- Reviews released: January 25, 2027
- Rebuttal period: January 25 to February 1, 2027
- Final decisions: February 25, 2027
AoE means Anywhere on Earth, UTC-12, the last timezone on the planet to reach midnight. That sounds generous until you remember the other half of the sentence: the submission deadline "will not be changed." There is no grace period to plan around, no extension rumor worth waiting for. The week between enrollment and submission is the whole runway.
Enrollment is the real first deadline
Submissions run through OpenReview, and the call requires an up-to-date OpenReview profile for every co-author. Here is the trap: new profiles created without an institutional email go through a moderation process that can take up to two weeks, while profiles created with an institutional email are activated automatically. Count backwards from November 10. A co-author who creates a profile in early November, from a personal email address, may still be waiting for moderation when enrollment closes. The fix is boring and immediate: every author creates or updates their profile this week, with an institutional email where one exists. The cvpr 2027 paper enrollment deadline will catch more teams than the submission deadline will, because it punishes paperwork rather than unfinished experiments.
Every author owes reviewer service
The call's Mandatory Reviewer Service section is short and absolute: all authors commit to serve as reviewers, area chairs, or senior area chairs when invited by the program chairs. Exceptions are granted only for truly exceptional circumstances, and those must be disclosed to the program chairs in advance and approved by them as valid reasons for exemption. Read that as part of the submission price. A ten-author paper is not just a paper anymore; it is a small reviewing workforce the conference expects to draw on. Labs submitting several papers should plan the reviewing load now, in October, rather than discovering it when invitations arrive in the middle of the rebuttal season. The cvpr 2027 reviewer requirement is not a suggestion box at the bottom of the form. It is a condition of playing.
The AI policy is still being written
The call has an LLM Usage section, and it is three sentences long. CVPR 2027 "will make use of large language models (LLMs) in selected parts of the conference and reviewing process." Then the honest part: "The exact scope, permitted uses, and associated guidelines are currently being finalized." The section will be updated with further details before the submission deadline. So the policy that governs how you may use these tools, and how the conference itself will use them in reviewing your paper, does not exist in final form roughly three weeks before the deadline. Nobody did anything wrong here; the field is moving and the policy is being written in public. But the practical consequence lands on authors. If your writing, coding, or experiment pipeline involves LLMs, check this section again in the final week before you submit, because the version you read today is explicitly a draft. The deadline will not move while the policy settles.
What a recorded deadline month can and cannot tell you
One worked example from our own records, because it teaches the right habit. This site's directory records CVPR as an A* tier conference with a tentative deadline month of November and a medium confidence label, last verified on September 15, 2026. This year, the recorded month matched: the official submission date is November 16. That is precisely what a recorded month is for. It tells you when to start watching a venue's own call; it never tells you the day a paper is due. The day only ever comes from the venue, which is why our month by month deadline guide treats every month as a prompt to verify, and why CVPR sits in our machine learning conferences comparison as a venue to plan around, not a date to trust secondhand.
One honest gap, stated plainly: the directory carries no recorded acceptance rate for CVPR. Only 8 of its 329 venues have a rate recorded, so this post quotes no odds for CVPR, because there is no recorded number to quote. If you are weighing CVPR against the rest of your season, that weighing is the job of Odds Reality Check: build the venue list, enter the rates you actually trust, and read the combined number before the season starts.
So the checklist for this week is short. Create the OpenReview profiles. Put November 10 and November 16 on the calendar as immovable. Budget the reviewing work your author list now owes. And reread the LLM section once more before you submit, because the call you are writing to is still, in that one respect, being written.