Motors · Avinox
Avinox · DJI Avinox (M2 series, flagship)

M2S

DJI's flagship is the most powerful production eMTB motor yet measured — 150 Nm Boost and a dyno-verified 1,450 watts at the cranks, in a 2.59 kg package. Here's what that actually feels like on the trail, where it lands against Bosch, and the one thing no-one can vouch for yet: how it holds up over time.

M2S eMTB motor
The DJI Avinox M2S — 150 Nm Boost and a measured 1,450 W from a 2.59 kg drive unit.
025050075010001250150060801001201401,450 Wcadence (rpm) →power (W)

Builds with a short, gentle ramp, peaks at 100-110 rpm and - unusually - holds power cleanly all the way to 150 rpm with no stalling at high cadence.

The verdict

Avinox M2S is the flagship full-power drive unit in the Avinox M2 series, launched 9 April 2026 and aimed squarely at the top of the full-power eMTB tier. The headline numbers: 150 Nm peak torque, 1500 W claimed peak (1462 W measured on an independent dyno), 36 V architecture, 2.63 kg. It supersedes the M1 (which measured around 850 W on the same test bench) and arrives as the highest-output production eMTB motor announced to date. Community framing is blunt: as one tester put it on a Forbidden Druid E build, the weight-to-power ratio is what stops people mid-sentence, with a 19.2 kg complete bike running well over a kilowatt on tap.

The numbers. Claimed peak is 1500 W and 150 Nm; eMTB Magazine's dyno landed at 1462 W, within 2.5% of the claim and well clear of anything else on the market. Nominal continuous output is 130 Nm and 1300 W, with the extra 20 Nm and 200 W reserved for the Boost burst, so the gap between cruising torque and the headline figure is real and worth understanding. Up to 800% assist is available in the top mode. Output is also battery-conditional: the FP700 700 Wh internal pack sustains the full 1500 W / 150 Nm, while the RS800 removable pack sustains 1300 W and bursts to 1500 W in Boost. That is still more sustained wattage than any rival's peak, so the RS800 is a feature ladder rather than a compromise. Firmware is delivered OTA through the Avinox app and has already shifted Trail-mode ramp behaviour twice since launch.

Character and feel. The recurring rider note is that the motor is quiet under load and clean off the pedals. Jazzii reports a 20.9 kg coil-shock build that is quiet but not TQ-silent, with no downhill rattle, which matches the wider consensus that drag is low enough to pedal over the 25 km/h cutoff without feeling like you are towing a brake. Modulation across Eco, Trail and Boost is where the Avinox app earns its keep: Rob Rides EMTB runs a beta app that lets him cap watts per mode, setting Trail to 600 W to mimic a Bosch CX feel and Turbo to 850 W to stretch range. Run uncapped, Boost is genuinely violent off a standstill and the 150 Nm shows up as front-wheel lift on steep tech.

Compatibility and ecosystem. Five battery configurations are supported: FP700 (700 Wh internal, full 1500 W sustained), RS800 (800 Wh removable, 1300 W sustained / 1500 W burst), RS600 (600 Wh removable, also rated to 1500 W peak), and FP800 / FP600 internal packs on non-PR builds capped at 1300 W. The Avinox app handles firmware, mode customisation and diagnostics. There is a bar remote and a top-tube display on most builds, with a 12 A / 508 W fast charger shipped on Amflow bikes. Standards are conventional: ISIS interface, Boost 148 rear, 55 mm chainline on most builds.

Reliability and known issues. Early-batch reports have been mostly clean. Mad_Angler1, a long-time Avinox brand watcher, flags that warranty politics are part of buying into the ecosystem and worth factoring in if you are not near a dealer. On the M1 generation, emtbeast's lab test showed the unit shutting down under sustained Boost load rather than hitting its claimed ceiling; M2S thermals appear improved on hairpin-wound stators, but long bike-park laps in 30 °C are still the real test and the jury is technically out until summer 2026 data lands. No formal recalls as of writing.

Bikes you'll find it on. The range now spans roughly sixty brands. For trail duty the Forbidden Druid E 2026 at 160/150 mm is the obvious benchmark, paired with the RS800 pack. The Amflow PX Carbon 2026 is the reference enduro build at 2.4 kg frame weight with the FP700 for full sustained output. For bigger terrain the Atherton S.170E 2026 runs 180/170 mm on the same drive unit, and the Crestline RS181 2026 pushes to 180/181 mm. Budget-leaning, the Megamo Reason 2027 brings the M2S into more sensible money. Tuning varies: Amflow's stock map is the most aggressive in Boost, while Megamo softens the initial ramp noticeably.

Verdict. The M2S suits riders who want the highest sustained output currently shipping and accept a 36 V ecosystem that is still maturing on the dealer-support side. The genuine trade-off is battery-conditional output: if you want the RS800's removability for shuttle days, you give up 200 W of sustained ceiling, and that choice is made at frame-purchase time, not later. Current flagship, launched 9 April 2026, no announced replacement.

“It's the weight-to-power ratio that stops people mid-sentence — a 19 kg bike with well over a kilowatt on tap.”

Sustained power & heat

How long the headline number actually lasts under sustained climbing load.

FP700 (700 Wh internal)
Holds 90% for 20 min · housing 120 °C

still 90%+ after 20 min continuous; housing past 120 C in an extreme bench scenario

Character

Rider input
Avinox publishes no fixed support-ratio figure. Assistance is set by named ride modes — Eco, Trail, Turbo and a 10–60 s Boost — each app-tunable per rider, rather than a single headline multiplier. In Boost the unit will return well over a kilowatt to roughly 100 W of rider input, but DJI does not quote a precise percentage.
On the trail
Minimal ramp-up and a predictable, almost calming delivery - it noticeably lowers the heart rate on technical climbs and salvages botched, steep lines on traction alone. Slightly less boisterous and nervous than some rivals. Against the M1 many riders already own, the M2S steps Boost peak torque from 120 Nm to 150 Nm and lifts sustained output to 130 Nm / up to 1500 W, while sealed bearings replace the M1 oil seal to cut unassisted drag (DJI launch figures, April 2026).
Noise
DJI's own lab figure is ≤45 dBA at full power (binaural mic, 0.1 m) — pin that to the not-yet-independent caveat. Remarkably quiet for its power class: close to light-assist motors in an alloy frame, a touch louder in carbon but still quieter than most full-power rivals. No rattle or clang.
Efficiency
Thirsty when you lean on it: the 700 Wh FP700 drains in well under an hour under sustained full Boost — call it roughly 12–15 Wh per km of hard climbing, so a long alpine day will want a spare or a charge stop. Ridden in Trail/Eco it is far more civil and competitive with lighter systems, but the power is there to be spent.

The case for and against

Strengths

  • The highest measured output of any production eMTB motor to date
  • Exceptional power-to-weight — just 2.59 kg
  • Holds power cleanly to 150 rpm, with no high-cadence stall
  • Remarkably quiet for its power class
  • Calm, predictable delivery that flatters technical climbs

Compromises

  • Thirsty: a 700 Wh pack empties in well under an hour at full boost
  • Runs hot under sustained load — housing past 120 °C on the bench
  • Trails Bosch CX and Specialized 3.1 on overrun/delay finesse
  • Too new to trust long-term: April 2026 launch means no real-world durability or reliability record yet

How it stacks up

The highest measured output of any production eMTB motor to date — 1450 W at the cranks against roughly 685 W from a Bosch CX Gen5 (100 Nm) and about 655 W from a Specialized 3.1 S-Works (111 Nm), all measured on the same lab dyno (Velomotion/ebike-lab) in a near-identical ~2.6 kg package. It trails Bosch CX and Specialized 3.1 on overrun/delay finesse — the Bosch eMTB+ mode feels more dynamic and intuitively tied to pedal force in tight technical sections — but nothing else matches its raw on-tap power-to-weight.
OWNER INTELLIGENCE
Most owners report no problems.
4Rattle / internal knock - on the motor marketed as rattle-free · typical onset: Ride 2, in three of the four cases.
4Launch-window parts & availability friction (system, not motor) · typical onset: Launch-window phenomenon.
2Speed-sensor fault -> motor replacement (the M1 pattern, repeating) · typical onset: Ride 2 / first weeks.
Read the full owner report →
Sources: Velomotion · ebike-lab.
Back
Top