How accurate is Envelope against real DOB filings?

98% of 1,375 DOB-approved New Building filings stayed within Envelope's maximum zoning floor area. On the 962 that built out the lot, Envelope was within 10% of the architect's filed figure 73% of the time. Comparable filings are DOB-approved New Building jobs that state a proposed zoning floor area, were examined under the rules Envelope applies today, and used at least half of the FAR DCP publishes for the lot. Every case is listed on runenvelope.com/accuracy with a link to the filing (engine 8e5ffcb9, October 17, 2026).

How is Envelope checked for accuracy?

Before every release, the engine is scored field by field against zoning studies by licensed NYC architects and engineers, to professional tolerances, and run against hundreds of DOB-approved New Building filings across every borough to confirm the city never approved more than the engine says is legal. A release that scores lower cannot ship.

What is Envelope checked against?

Zoning studies by practicing NYC architects and engineers on real lots, and DOB-approved New Building filings stratified by borough, district family, special district, waterfront, corner and split-zone conditions.

Can a release lower the accuracy score?

No. An accuracy gate runs on every deployment and blocks any release whose score drops.

Does Envelope replace an architect?

No. It replaces the first $1,500 to $10,000 and the weeks of waiting: the study you run to decide whether a lot deserves the architect.

Method

Anyone can tell you the FAR. Will it survive a DOB examiner?

98% of 1,375 DOB-approved New Building filings stayed within Envelope's maximum zoning floor area. On the 962 that built out the lot, Envelope was within 10% of the architect's filed figure 73% of the time.

Benchmark

Envelope against what architects filed with DOB

Every DOB New Building filing states the architect's proposed zoning floor area, and an approval is the city's acceptance that this much floor area is legal on the lot. We ran Envelope on the same tax lot, through the same engine path a report uses, for every comparable approved filing, and published every result below with a link to the filing.

98%Filings within Envelope's maximum1,351 of 1,375 approved filings did not exceed the maximum zoning floor area Envelope reports for their use (5% tolerance). 24 did.

73%Built-out filings within 10%706 of 962 filings that used at least 90% of DCP's published FAR landed within 10% of Envelope's maximum. Median gap +0.3%.

1,375Lots scoredOut of 13,975 approved New Building filings; 8,242 were set aside by the rules below and 1 could not be run.

ENGINE 8e5ffcb9 · AS OF OCTOBER 17, 2026 · SOURCE NYC OPEN DATA IC3T-WCY2 · REFRESHED WEEKLY

1. Did the city approve more than Envelope allows?

On every scored filing: is the filed zoning floor area at or under Envelope's maximum for the lot, plus 5% for cellar, rounding and survey differences? A filing above it means the city approved floor area Envelope says is not there, an Envelope under-report. A filing below proves little on its own, because most buildings do not max out. That is why there is a second test.

Filings with dwelling units are compared with Envelope's residential maximum, or with its UAP / MIH maximum where the filing reaches into it (those programs are as of right and Envelope reports them). Filings with no dwelling units (community facility, commercial) are compared with the highest maximum Envelope reports for any use. So are transient hotels: BIS lists a hotel's rooms as dwelling units, but hotel floor area is commercial. In R6 to R10 districts without a letter suffix the residential maximum includes Envelope's sky exposure plane building (ZR 23-732), the optional form that reaches 2.43 FAR in R6, for the containment test only; the built-out gap below is measured against the standard ZR 23-22 maximum, the form architects actually use on most of these lots. Against the standard residential maximum alone, without UAP or MIH, 87.2% of filings fit; most filings approved before City of Yes could not use UAP, so the gap between the two figures is worth reading in the case list.

A residential filing above every residential maximum but under Envelope's community facility or commercial maximum counts against Envelope. DOB filings do not split floor area by use, so such a building is either part community facility or a residential figure Envelope set too low, and we do not get to assume the first. The exception is a filing whose building has since been built: where MapPLUTO dates the building to the filing and its non-residential share puts the residential floor area within Envelope's residential maximum, the filing counts as contained, in its own bucket ( "Mixed, built" in the case list). MapPLUTO's floor-area split is an assessor's estimate, so this is evidence rather than a measurement; an all-residential building stays against Envelope. Counting them as passes would raise the figure to 99.7%; 20 filings are in that bucket.

2. When the architect built out, how close was Envelope?

The second test only looks at filings whose floor area reaches at least 90% of the FAR that the Department of City Planning publishes for the lot in MapPLUTO. These are the architects who tried to use everything. Here an Envelope over-statement would show, so we report how often Envelope's maximum lands within 5%, 10% and 20% of the filed figure, and in which direction it misses.

Which filings count as built out is decided by DCP's number, never by Envelope's, so the cohort cannot be chosen to flatter the engine. Nothing on this page is tuned after the fact: the rules are fixed in code and applied before Envelope runs.

Within 5%: 62%
Within 10%: 73%
Within 20%: 85%
Envelope higher by 10%+: 99
Envelope lower by 10%+: 157
Median absolute gap: 2.1%

What was left out, and why

A filing is only a fair test of an as-of-right maximum if the architect was working under the same rules, on the same lot, without a bonus or a variance. Each rule below uses only the filing and MapPLUTO, never Envelope's answer, and each filing is counted under the first rule it meets.

Rule	Why it is not comparable	Filings
Approved New Building filings with a stated ZFA	BIS primary filings approved from 2017 through 2026. DOB NOW filings publish no zoning floor area and cannot be scored.	13,975
Withdrawn filing	BIS withdrawal flag set: the job was withdrawn, not built under this filing.	−616
Not approved	Job status is not plan-exam approved, permitted or signed off.	−1
Condo or malformed lot	The filing names a condominium unit lot or an unusable block/lot, so there is no single tax lot to run.	−275
Superseded by a later filing	A later approved New Building filing exists on the same lot; only the latest is scored.	−1,134
Lot no longer exists	The filed tax lot is not in today's MapPLUTO: it was merged, apportioned or condominiumized after the filing, so today's lot is not the lot that was filed.	−785
No zoning district	MapPLUTO carries no zoning district for the lot (park, street bed, unmapped).	−28
Landmark	The filing is flagged landmarked: LPC review can shape or cap the building independently of zoning.	−48
Discretionary action named	The job description names a variance, special permit, CPC action, zoning lot merger, transferred development rights or a floor-area bonus.	−13
Rezoned since the filing	The district on the filing is not a district on the lot today, so the filing was examined under different rules than the ones Envelope applies.	−239
R1 to R5, approved before City of Yes	City of Yes for Housing Opportunity (adopted 2024-12-05) rewrote low-density floor area (ZR 23-21: attic and infill allowances gone, new columns). A pre-adoption R1 to R5 approval (including C3, C3A and C4-1, their commercial equivalents) was examined under a maximum that no longer exists.	−4,014
Under 2,500 sf	A filed zoning floor area under 2,500 sf is a garage, shed or foundation-only job and says nothing about a lot's envelope.	−111
No lot area	MapPLUTO has no lot area, so implied FAR cannot be computed.	−2
No published FAR for the use	DCP publishes no FAR for the filing's use on this lot (for example a residential filing where MapPLUTO's residential FAR is zero): a use variance, a special district or a data gap.	−49
Multi-lot project total	The filed floor area exceeds the filed stories times the whole lot (physically impossible on this lot), or the identical floor area is filed on two or more lots of the same block: one project's total booked against each lot.	−359
Building spans several lots	The standing building's average floor plate (MapPLUTO floor area / floors) is larger than the tax lot, so the zoning lot is bigger than the tax lot.	−38
Zoning lot is not the tax lot (ACRIS)	ACRIS records, between six years before and three years after the approval, a development-rights or air-rights conveyance on the lot, or a zoning lot description that also names another tax lot that still exists: the floor area was filed for a zoning lot (ZR 12-10) that is not the tax lot Envelope runs, in whichever direction the floor area moved.	−193
Zoning map amended since the approval	An adopted zoning map amendment (DCP dcp_zoning_map_amendments) covering at least 10% of the lot took effect after the approval, so the filing was examined under a map that no longer applies even where the district name survived (for example a new special district or MIH area).	−7
Above DCP's published maximum	The filing's implied FAR is more than 5% above every FAR DCP publishes for the lot. Only a bonus, transferred floor area, a merger or a variance reaches that, none of which is as of right.	−144
Used under half the FAR	The filing's implied FAR is under half of DCP's published FAR for its use. The architect did not try to build out, so the filing cannot test a maximum.	−186
Comparable filings, run through Envelope	1 could not be run by the engine and are reported as failures, not dropped silently.	1,376

The biggest exclusion is low-density: almost every BIS New Building approval in R1 to R5 predates City of Yes, which rewrote floor area in those districts, so this benchmark does not yet speak to them. Filings were examined under the rules in force when they were approved; Envelope applies today's, which is why a filing on a lot rezoned since is set aside.

By borough

	Filings	Within max	Above max	Built out	Within 10%	Median gap
Brooklyn	707	98.9%	8	497	78%	+0.2%
Bronx	366	99.2%	3	233	64%	+0.4%
Queens	187	95.2%	9	152	78%	+0.2%
Manhattan	95	95.8%	4	68	74%	+0.3%
Staten Island	20	100.0%	0	12	25%	-2.6%

By district family

	Filings	Within max	Above max	Built out	Within 10%	Median gap
R6 and R7 (medium density)	1,106	98.5%	17	793	75%	+0.2%
Commercial (C1 to C8)	117	97.4%	3	83	55%	+0.5%
R8 to R10 (high density)	56	98.2%	1	28	71%	+1.7%
Manufacturing (M1 to M3)	47	97.9%	1	25	100%	+0.4%
Mixed use (MX, M1 with R)	43	95.3%	2	33	73%	+0.3%
R1 to R5 (low density)	6	100.0%	0	0	n/a	n/a

By zoning district (8 or more filings)

	Filings	Within max	Above max	Built out	Within 10%	Median gap
R6	480	97.5%	12	335	59%	+0.1%
R6B	192	100.0%	0	165	97%	+0.5%
R7-1	170	99.4%	1	121	67%	+0.3%
R6A	124	100.0%	0	88	98%	+0.6%
R7A	74	100.0%	0	43	93%	+0.4%
R8	44	97.7%	1	20	60%	+4.3%
R7-2	40	90.0%	4	30	47%	-0.9%
C4-2	20	95.0%	1	17	12%	+3.0%
C4-3	18	94.4%	1	18	28%	-4.8%
M1-2/R6A	14	100.0%	0	10	90%	+0.9%
M3-1	10	100.0%	0	5	100%	0.0%
R7D	10	100.0%	0	0	n/a	n/a
M1-2	9	100.0%	0	2	100%	+4.0%
R7B	9	100.0%	0	7	100%	+0.6%
C8-1	8	100.0%	0	4	100%	+1.7%
M2-1	8	100.0%	0	5	100%	+0.4%

By approval date

	Filings	Within max	Above max	Built out	Within 10%	Median gap
2017-2019	756	98.7%	10	549	75%	+0.2%
2020-2024 (before City of Yes)	599	97.8%	13	406	71%	+0.3%
After City of Yes	20	95.0%	1	7	57%	+0.1%

Special districts and split lots

	Filings	Within max	Above max	Built out	Within 10%	Median gap
No special district	1,242	98.6%	18	869	74%	+0.3%
Special district	133	95.5%	6	93	71%	+0.3%
Single district	1,314	98.6%	19	931	74%	+0.3%
Split lot	61	91.8%	5	31	55%	-5.4%

Median gap is Envelope's maximum against the filed figure on built-out filings: positive means Envelope allows more than the architect filed, negative means less.

Filed ZFA divided by Envelope's maximum

0.00 to 0.50

0.50 to 0.70

112

0.70 to 0.90

283

0.90 to 1.00

672

1.00 to 1.05

108

1.05 to 1.20

1.20 to 1.50

1.50 and up

Yellow bars are filings at or under Envelope's maximum (plus the 5% tolerance); black bars are filings the city approved above it.

Every case

Each job number opens the filing on DOB's Building Information System, so you can check the filed figure yourself. Filed / max is the filed zoning floor area divided by Envelope's maximum for the filing's use; above 1.05 the city approved more than Envelope allows.

BoroughVerdictSortAddress or job numberBuilt-out filings only

1,375 filings

DOB job	Address	District	Approved	Lot sf	Filed ZFA	Envelope max	Filed / max	Verdict
210182354	215 E 163Rd Street, Bronx BBL 2024550044	R7-1	2026-05-08	5,333	26,198 FAR 4.91	18,346 FAR 3.44	1.43 built out	Within UAP / MIH
540177093	1 Ballard Avenue, Staten Island BBL 5060460003	R3X SRD	2026-03-12	5,829	2,503 FAR 0.43	4,372 FAR 0.75	0.57	Within
520368405	9 Inez Street, Staten Island BBL 5067050096	R3X SRD	2026-01-27	5,500	3,279 FAR 0.60	5,500 FAR 1.00	0.60	Within
321597660	781 Livonia Avenue, Brooklyn BBL 3040740039	R5	2026-01-21	2,500	3,051 FAR 1.22	3,750 FAR 1.50	0.81	Within
321589474	739 Blake Ave, Brooklyn BBL 3037750150	R6	2025-12-12	4,000	8,442 FAR 2.11	8,800 FAR 2.43	0.96	Within
321597802	943 Dumont Ave, Brooklyn BBL 3040630052	R5	2025-12-12	5,400	6,714 FAR 1.24	8,100 FAR 1.50	0.83	Within
220212641	672 St Anns Ave, Bronx BBL 2026170020	R6	2025-12-05	2,589	7,201 FAR 2.78	7,767 FAR 3.00	0.93 built out	Within
220212650	901 Eagle Ave, Bronx BBL 2026200046	R6	2025-12-05	2,500	5,481 FAR 2.19	5,500 FAR 2.43	1.00 built out	Within
220212669	675 Eagle Ave, Bronx BBL 2026170070	R6	2025-12-05	1,931	4,110 FAR 2.13	5,793 FAR 3.00	0.71	Within
220212696	1048 Faile Street, Bronx BBL 2027480024	R7-1	2025-12-05	2,050	4,743 FAR 2.31	7,052 FAR 3.44	0.67	Within
220212703	959 Home Street, Bronx BBL 2029790001	R7-1	2025-12-05	3,111	10,802 FAR 3.47	12,444 FAR 4.00	0.87 built out	Within
220212712	1298 Hoe Avenue, Bronx BBL 2029870014	R7-1	2025-12-05	2,500	6,376 FAR 2.55	8,600 FAR 3.44	0.74	Within
220212730	1013 Home Street, Bronx BBL 2029930033	R7-1	2025-12-05	2,143	5,469 FAR 2.55	8,572 FAR 4.00	0.64	Within
220212785	1109 Intervale Avenue, Bronx BBL 2026920073	R7-1	2025-12-05	2,283	5,713 FAR 2.50	9,132 FAR 4.00	0.63	Within
220212678	667 Cauldwell Ave, Bronx BBL 2026240073	R6	2025-12-04	2,875	6,320 FAR 2.20	6,325 FAR 2.43	1.00 built out	Within
220212758	881 Brook Ave, Bronx BBL 2023650023	C4-4	2025-11-12	1,764	4,121 FAR 2.34	7,056 FAR 4.00	0.58	Within
520322963	85 Santina Drive, Staten Island BBL 5065170097	R3X SRD	2025-08-13	6,908	2,875 FAR 0.42	5,181 FAR 0.75	0.55	Within
520303742	42 Santina Drive, Staten Island BBL 5065170087	R3X SRD	2025-07-30	6,470	3,993 FAR 0.62	4,852 FAR 0.75	0.82	Within
321596858	1211 Fulton St, Brooklyn BBL 3018420066	C4-5D	2025-04-04	5,456	25,874 FAR 4.74	25,425 FAR 4.66	1.02 built out	Within
210180221	50 Willis Avenue, Bronx BBL 2022600180	M1-5/R8A / M3-1 MX-1	2025-02-11	29,530	175,935 FAR 5.96	129,341 FAR 4.38	1.36 built out	Over
321601520	372 Livonia Avenuee, Brooklyn BBL 3038120019	R7-2	2024-12-04	9,150	55,505 FAR 6.07	31,476 FAR 3.44	1.76 built out	Over res.
210179876	2305 Belmont Avenue, Bronx BBL 2030870056	R7-1	2024-09-09	2,500	8,544 FAR 3.42	8,600 FAR 3.44	0.99 built out	Within
321383793	4908 8 Avenue, Brooklyn BBL 3007860039	R6	2024-06-05	8,013	37,248 FAR 4.65	24,039 FAR 3.00	1.55 built out	Over res.
321590845	329 Van Siclen Ave, Brooklyn BBL 3040100006	R6	2024-05-28	2,500	5,132 FAR 2.05	5,500 FAR 2.43	0.93	Within
321597214	342 Van Siclen Avenue, Brooklyn BBL 3040090047	R6	2024-05-21	2,090	4,199 FAR 2.01	4,598 FAR 2.20	0.91	Within

1 to 25 of 1,375

Against licensed professionals' studies

Envelope keeps a library of zoning studies prepared by practicing architects and engineers for real New York lots, and scores the engine against every figure they state:

■FAR by use (residential, UAP or qualifying, community facility, commercial)
■Maximum zoning floor area under each
■Base height and maximum height, per portion on split lots
■Rear yard, rear yard above 75 ft, setback
■Lot coverage, per portion on assemblages
■Dwelling unit maximums
■Site conditions: corner, wide street, waterfront block, flood zone, Inclusionary Housing area, split zoning, landmark

Tolerances. FAR to two decimal places. Floor area to a few percent, after normalizing to the study's surveyed lot area, because PLUTO and a survey rarely agree to the square foot. Heights, yards, setbacks and site conditions: exact.

The library spans assemblages, corner lots, split-zoned lots, waterfront blocks, special districts and mixed-use districts, because those are the conditions where a generic FAR lookup fails and where a study is expensive. Studies written under rules City of Yes has since rewritten stay in the library; they are scored against the rules in force when they were written.

Against what the city actually approved

A study is one professional's opinion. A DOB-approved New Building filing is the city's. Beyond the published benchmark above, a stratified set of approved filings (BIS and DOB NOW) is checked on every release across all five boroughs by district family (R1 to R5, R6 to R7, R8 and up, C, M, MX), special district, waterfront block, corner and split zone, on height and dwelling units as well as floor area.

For each lot one question is asked: did the city approve more than the engine says is legal? A filing above it is a defect unless it is explained by a bonus the engine itself models (UAP, MIH, IH), a discretionary action (a CPC special permit, a landmark transfer), a multi-building project total booked on one lot, or an approval under rules that have since changed. Every verdict is logged with its reason, and every unexplained over-approval is tracked to a root cause.

Before every deploy

Both checks run as a gate on every deployment, alongside the full test suite. The gate records the score; if a release would lower it (a professional field that no longer matches, a new unexplained over-approval, a height or base that moved on a lot that used to be clean), the release is blocked rather than warned about. A data source outage fails the run instead of lowering the bar.

What this does not claim

Envelope models what is permitted as of right. It does not promise what a special permit, variance, certification or rezoning might yield; the report says so rather than quietly adding the floor area. An approved filing is evidence of what the city accepted, not a survey: the benchmark uses MapPLUTO lot areas, and a filing that stays under Envelope's maximum does not prove the maximum is right, which is why the built-out test exists. Envelope is not a licensed architect, engineer, surveyor or attorney, and a report is not professional advice. It is a very fast, very well-cited first read, designed to tell you which lots deserve the expensive version.

FAQ

Questions people ask

How accurate is Envelope against real DOB filings?: 98% of 1,375 DOB-approved New Building filings stayed within Envelope's maximum zoning floor area. On the 962 that built out the lot, Envelope was within 10% of the architect's filed figure 73% of the time. Comparable filings are DOB-approved New Building jobs that state a proposed zoning floor area, were examined under the rules Envelope applies today, and used at least half of the FAR DCP publishes for the lot. Every case is listed on runenvelope.com/accuracy with a link to the filing (engine 8e5ffcb9, October 17, 2026).
How is Envelope checked for accuracy?: Before every release, the engine is scored field by field against zoning studies by licensed NYC architects and engineers, to professional tolerances, and run against hundreds of DOB-approved New Building filings across every borough to confirm the city never approved more than the engine says is legal. A release that scores lower cannot ship.
What is Envelope checked against?: Zoning studies by practicing NYC architects and engineers on real lots, and DOB-approved New Building filings stratified by borough, district family, special district, waterfront, corner and split-zone conditions.
Can a release lower the accuracy score?: No. An accuracy gate runs on every deployment and blocks any release whose score drops.
Does Envelope replace an architect?: No. It replaces the first $1,500 to $10,000 and the weeks of waiting: the study you run to decide whether a lot deserves the architect.

Run it on the lot you are looking at today.

Run a report See a sample first