PERM processing time tracker: where DOL's queue is now, and every update in 2026

When DOL's queue reached each filing month in 2026

DOL's page only shows where the queue is today. This table keeps when it got to each filing month, for applications with no audit (Analyst Review) and audited ones (Audit Review). Each cell names the earliest DOL update we have read that showed the queue at that month or later. "Between" gives two archive days: the last copy that still showed an earlier month, and the first copy that showed this month or later. The queue got there between those two days. That holds even for days we haven't read yet. But an unread update in between may have reached the month first, so the stamp is the earliest we have read, not always the first one DOL posted. Archive days are not DOL's posting days. The 1 old copy of the page is left out.

Download: perm-filing-month-reached-2026.csv. A month the queue reached means DOL started on that month's filings. It doesn't mean those applications were decided.

DOL's current PERM figures

The Department of Labor posts one PERM line and overwrites it each time it changes. This page keeps every version we could read from the Wayback Machine's daily copies of flag.dol.gov/processingtimes.

As of DOL's update stamped 10/05/2026 (first in the archive on Oct 6, 2026), DOL is working on PERM applications filed in December 2025 (Analyst Review) and December 2025 (Audit Review). The average for PERM applications decided in August 2026 was 336 calendar days.

Newest copy read: Oct 6, 2026. DOL may have updated since. The live figure is at flag.dol.gov/processingtimes.

DOL's own explanation: the queue months "reflect the month and year in which cases were filed and are currently being adjudicated", and "OFLC may be completing the processing of applications filed prior to the month posted". The average days figure covers "all PERM applications for the most recent month".

Every version of DOL's PERM figures we could read in 2026

Newest first. "First seen" and "last seen" are the days the Wayback Machine copied the page, not the days DOL posted. Each change happened somewhere between one version's last-seen day and the next version's first-seen day. Average days are shown with the month DOL says they cover, as printed ("N/A" is DOL's); "blank" means no figure, or "--", was printed; "not shown" means the row had no month.

Download the table: perm-versions-2026.csv.

What the history shows

• 19 different "as of" stamps in 22 versions between Jan 2, 2026 and Oct 6, 2026. DOL's update schedule says "PERM: Updated COB at end of first work week of each month"; in the days read, the page changed more often than that. 13 of the 21 changes came later in a month: the old version was still up on or after the 8th and the new one appeared before the month ended. This is a minimum: days not read yet can hold more versions.
• One stamp had the wrong year. The version first seen Jan 8, 2026 was stamped "01/05/2025", about a year before it appeared. If you sort DOL's updates by their stamps, this one lands in the wrong year.
• 3 updates changed only the stamp (01/05/2026, 8/20/2026, 8/31/2026): a new date, the same figures.
• 3 updates changed figures without changing the stamp (stamp 03/12/2026, 8/7/2026, 8/28/2026). The stamp alone doesn't tell you the page changed.
• The archived copy for Jul 9, 2026 is an old page. Its PERM section is word for word the 04/23/2026 version, which had already been replaced. Read on its own, it would look like the queue moved back months. We list it in the table and leave it out of everything else. We don't know whether DOL's site or the archive served the old copy.
• Audit Review moved back: from Dec 2025 (stamp 04/23/2026) to Nov 2025 (stamp 04/30/2026). The queue month doesn't only go forward.

We don't print a speed for the queue, or a date for your case. The rows above are what DOL posted, and nothing more.

How much of 2026 this covers

The Wayback Machine has a copy of DOL's page for 274 of the 279 days from Jan 1, 2026 to Oct 6, 2026. It has no copy for Jan 16, Mar 3, Jul 18, Aug 16, Sep 26. We read one copy per day, starting with every 8th day across the year and then filling in. Days read so far: 167 of 274 (every 8th day: 35 of 35; every 4th day: 54 of 69; every 2nd day: 94 of 137; every day: 167 of 274). The longest run of archived days not read yet is 4 days (Jan 14 to Jan 17), so a version that was up for less than 4 days could be missing from the table.

Each copy is read by the same program (tools/perm-captures.py), which takes the PERM section of the page as printed and groups days whose text is identical. The raw first copy of each version is kept in the project's repository. This is not legal advice, and we are not affiliated with DOL.