Head-to-head comparisons

Triton ULTRA vs.

HALO

The closest feature match. The differences show up in occupancy (built in vs. model- and accessory-dependent), daisy-chaining (5 sensors vs. 2), warranty (10 years without cloud vs. 5) and software, where Triton ULTRA offers a more modern interface and better notification options,including SMS texts.

Read the full comparison

Triton ULTRA vs.

Verkada

A general-purpose air quality sensor inside Verkada’s cloud platform. A paid license is required for every sensor, and there’s no people counting, keyword detection or THC identification.

Read the full comparison

Triton ULTRA vs.

Zeptive

The battery-powered option. Easy to mount without cable, but it brings battery swaps about every three months, a 140 Wh lithium-ion pack in the ceiling and a narrower safety feature set.

Read the full comparison

Triton ULTRA vs.

Axis

A well-built general-purpose air quality sensor that fits naturally in Axis camera systems. It’s hard to manage in bulk, with settings and alerts spread across several Axis products instead of one cloud interface, and it lacks features designed for schools, like people counting and keyword detection.

Read the full comparison

Triton ULTRA vs.

FlySense

A hardwired vape and smoke sensor with sound-based aggression alerts. No THC detection, occupancy, keywords, published warranty or published integrations.

Read the full comparison

All six, side by side

Based on each vendor’s public documentation. Features can vary by model, firmware and license,
so confirm the exact configuration you’re quoted.

Included

Limited, model-specific or add-on

Not offered or not documented

Capability

Triton ULTRA

HALO

HALO Smart Sensor 3C / 4

Full comparison

Verkada

SV23 / SV25

Full comparison

Zeptive

ZVD2200 / ZVD2300

Full comparison

Axis

AXIS D6310

Full comparison

FlySense

Soter FS300

Full comparison

detection

Vape and THC detection

Included. Nicotine and THC vape pens, plus separate THC detection for combusted cannabis

Vape detection with separate THC detection

Vape Index (0–100) for suspected vaping or smoking; no THC identification

Nicotine, non-nicotine and THC vaping, plus smoke

Vaping and smoking; no separate THC classification documented

Limited or add-on.Vape and cigarette smoke; no separate THC detection

Published accuracy testing

98.9% overall accuracy across 8,400 events, including perfumes and deodorants

No comparable published test found

No published test found

Restroom safety beyond vaping

People counting

Built-in radar people counting

Not built in as standard; requires the separate HALO 3C-PC or HALO 4 Occupancy Counter model

Not on the sensor; needs a Verkada camera with People Analytics, which can’t go in a restroom

Not supported; PIR motion sensing only

PIR presence only (about 5 m); no count

Not supported

Position, dwell time and loitering

Patented radar view of each person’s approximate position and dwell time, with loitering alerts

Not built into the sensor; loitering requires Amplify, a separate add-on hardware device

Not included

Prolonged-presence alerts only (PIR)

Not supported

Spoken emergency keywords

Standard keywords; custom words and phrases with Triton Cloud

Spoken keyword alerts; customization not documented

Not included

Detects that speech occurred, not what was said

Not included

Aggression and audio events

Aggression and sound-level alerts; AI classification for glass break and screams/ shouts; optional gunshot detection

Aggression and gunshot detection; HALO 4 adds scream, glass break, loud noise, smoke and car alarms, metal crash and other audio analytics

Noise-level readings and thresholds

Decibel-based aggression and noise alerts (optional sound meter)

Classifies scream, shout, speech, glass break and coughing fit

Aggression alerts from sound frequency, decibel level and patterns

Cell phone detection

Included

Not supported

In-room response

RGB light ring and speaker for pre-recorded or custom messages

Light and speaker for pre-recorded or custom messages; HALO 4 adds two-way communication (live listening in restrooms raises privacy questions)

Not included

Audible buzzer

LEDs, 4 W speaker and two-way audio over SIP (live listening in restrooms raises privacy questions)

LED status; buzzer for tampering only

Environmental monitoring

PM, TVOCs, formaldehyde, CO, CO2, NO2, ethanol, temperature, humidity, humidex

Air-quality and environmental readings

Air-quality readings; SV25 adds CO, formaldehyde, light and pressure

VOCs, particulates, temperature and humidity

CO2, PM1.0–PM10, VOC and NOx index, AQI, temperature, humidity

VOCs, temperature and humidity

Software and alerts

Works without a subscription

Yes. Local software included; Triton Cloud optional

Yes. Local operation; HALO Cloud optional

No. Cloud only, with a paid license per sensor renewed by term

No. Cloud only; no local software documented

No sensor license; dashboards (AXIS Data Insights) licensed per device

Cloud portal (EoT Live) with subscription; no local software documented

Integrations

25+ platforms across video, access control, intercom and mass notification; live occupancy view streams to your VMS over RTSP

VMS integrations and dashboard streams; licenses vary by model

Native in Verkada Command; alert events only to 4 third-party VMSs

10 video platforms; alerts arrive as VMS events

Open standards (VAPIX, MQTT, ONVIF, SIP, Modbus TCP); pairs natively with Axis devices

No specific integrations published

Incident review with camera footage

Evidence Room pairs each alert with the matching camera footage automatically

Through supported security integrations

Side by side with Verkada cameras in Command (each camera needs its own license)

Alerts arrive as events in your VMS

Through AXIS Camera Station or a third-party VMS

Not documented

Alert delivery

Email locally; SMS and app push with Triton Cloud; Microsoft Teams, Slack and Google Chat

Email and cloud mobile alerts; SMS described as beta, on request

SMS, email, mobile app and webhooks

Email, SMS and app push

Email, HTTP, MQTT, SNMP and SIP; phone alerts depend on your VMS

Portal, email and SMS

Installation, warranty and long-term cost

Power

PoE or PoE+ (12V DC with Wi-Fi also supported); no battery

PoE or 12V DC

PoE

ZVD2200: PoE or USB. ZVD2300: 140.4 Wh lithium-ion battery, about 3 months per charge

PoE

PoE (802.3af)

PoE daisy-chaining

Up to 5 sensors on one PoE+ line

2 sensors with PoE+

One cable per sensor

One cable per wired sensor

Passthrough to one additional device

One cable per sensor

Standard hardware warranty

10 years, no cloud subscription required

5 years; 10 years with a 10-year cloud software purchase

10 years (paid license required to operate)

5 years; ends if committed subscription fees go unpaid; batteries 12 months

5 years

Not published

What matters most, and how the field compares

The five questions school safety teams tell us decide the purchase,
with an honest read of where each product stands.

Can staff trust the alerts?

Every vape detector looks similar on a spec sheet. What decides whether staff keep responding is how often an alert is real, and whether the sensor ignores body spray, perfume and cleaning products. A detector that cries wolf gets muted.

Ask every vendor for published testing. Triton’s published test detected 2,861 of 2,900 vape events and correctly rejected 5,454 of 5,500 non-vape events, for 98.9% overall accuracy across 8,400 events. That’s Triton’s own test, not a head-to-head result, so run a side-by-side trial in your own building too.

THC matters as well. Schools often respond differently to nicotine and THC, so check whether a sensor identifies THC separately or just reports “vaping or smoking.”

Where the field lands

Triton ULTRA is the only product here with published accuracy testing we found. Triton ULTRA, HALO and Zeptive identify THC; Verkada, Axis and FlySense do not document separate THC identification.

What extra value does the sensor provide?

Restrooms and locker rooms are where cameras can’t go and staff can’t supervise, which is why students choose them for vaping, fights and bullying. This is where the products differ most.

Triton ULTRA’s patented radar occupancy (U.S. Patent No. 12,290,365) counts people, shows each person’s approximate position and dwell time, and alerts on extended stays, without a camera. Staff can see whether a restroom is empty, crowded or has one student who has stayed too long before anyone walks in.

Most other sensors use passive infrared (PIR), which detects motion or presence but can’t count or locate people. Spoken keyword alerts let a student call for help, and cell phone detection flags phone use in rooms where it isn’t allowed.

Where the field lands

Triton ULTRA includes counting, positions, dwell time and loitering in every sensor. HALO offers counting on specific models and loitering through an accessory. Zeptive offers prolonged-presence alerts. Verkada, Axis and FlySense don’t count people. Keyword alerts: Triton ULTRA and HALO. Cell phone detection: Triton ULTRA only.

Who controls the software?

There are three software models in this market, and they shape your costs for the life of the system.

Local software with optional cloud (Triton ULTRA, HALO): the sensors keep working and alerting without a subscription. Cloud adds features like SMS, app alerts and district-wide reporting.

Cloud required (Verkada, Zeptive, FlySense): a paid license is part of the cost of operating the sensor. Ask what stops working if a renewal lapses. Verkada notifications end 30 days after licenses expire, and Zeptive’s warranty ends if subscription fees go unpaid.

A toolkit (Axis): no license on the sensor itself, but alerts, management and dashboards are spread across separate Axis products or your VMS.

Where the field lands

Triton ULTRA and HALO are the two options that run without a required subscription.

Will it work with what you already run?

A sensor alert is only useful if it reaches the right person and is easy to investigate. Check how alerts get to administrators’ phones, and how long it takes to find the matching camera footage.

Triton ULTRA integrates with more than 25 platforms, streams its live occupancy view into your VMS, and its Evidence Room pairs each alert with the matching camera footage so staff don’t have to scrub through video. If your district is standardized on Verkada or Axis cameras, their own sensors pair natively; Triton ULTRA integrates with Verkada Command and AXIS Camera Station as well.

Where the field lands

Triton ULTRA, HALO and Axis have the broadest integration options. Verkada works best inside its own platform. Zeptive sends alerts to 10 video platforms. FlySense doesn’t publish an integration list.

Frequently Asked Questions

Which vape detector is the most accurate?

Triton publishes 98.9% overall accuracy across 8,400 test events, including rejection of perfumes, deodorants and other everyday sources. We found no comparable published testing from the other vendors reviewed here. Because performance depends on placement and airflow, test any finalists side by side in your own building.

Which vape detectors can count people in a restroom?

Triton ULTRA counts people with built-in radar and shows each person’s approximate position and dwell time. HALO offers counting on its 3C-PC and HALO 4 Occupancy Counter models. Verkada, Zeptive, Axis and FlySense sensors don’t count people; several use PIR motion or presence sensing instead.

Do vape detectors require a subscription?

It depends on the product. Triton ULTRA and HALO include local software that works without a subscription, with optional cloud services. Verkada, Zeptive and FlySense are managed through cloud platforms that require a paid license or subscription.

Which vape detector has the longest warranty?

Triton ULTRA and Verkada both carry 10-year standard hardware warranties; Triton ULTRA’s doesn’t require a cloud subscription. HALO is 5 years standard (10 with a 10-year cloud purchase), Zeptive and Axis are 5 years, and FlySense doesn’t publish its terms.

Is a battery-powered vape detector a good idea for schools?

For trials, temporary placements or rooms where cable can’t be run, it can be. For permanent installations, factor in battery changes about every three months per sensor and a large lithium-ion battery mounted in the ceiling.

How Triton ULTRA compares to HALO, Verkada, Zeptive, Axis and FlySense

The short version

Head-to-head comparisons

All six, side by side

What matters most, and how the field compares

Can staff trust the alerts?

What extra value does the sensor provide?

Who controls the software?

Will it work with what you already run?

When another option might fit

Zeptive

Verkada

How to run a fair side-by-side trial

Compare them in your own school

Frequently Asked Questions

Sources