NAM A2 Native — Any Amp You Want

Firmware update · Available now

ANY AMPYOU WANT.NAM A2 NATIVE

Load real Neural Amp Modeler captures straight onto your NUX pedal. No conversion. No proprietary re-encoding. The same open file the community trains — running on your board.

NUX hardware runs NAM A2-Lite natively: the embedded-optimised size packed inside every A2 file.

The basics

What is NAM?

Neural Amp Modeler is an open-source way of capturing gear. Instead of writing equations that describe a circuit, you record what a real amp actually does to a signal — and a neural network learns the relationship between what went in and what came out.

The result is a .nam file: a model of that amp, at that setting, with its harmonic content, dynamics and response under your hands intact. It can capture amps, cabs, full rigs, drive pedals and outboard gear.

Because NAM is open source, a .nam file isn’t locked to any brand. It behaves the way an IR or a MIDI file does — make it once, play it anywhere NAM is supported. TONE3000 hosts a library of them, free to download.

Real ampor pedal, or rig Record itdry in, wet out Trainnetwork learns it .nam fileopen format Your NUXAMP block CAPTURE → TRAIN → PLAY
A capture is made once, then plays anywhere NAM is supported.

The difference

What “native” means

There are two ways to get a NAM capture running on a guitar pedal. They are not the same thing, and the distinction is worth understanding before you buy anything — from us or from anyone else.

Converted. The device reads the .nam file, re-encodes it into the manufacturer’s own internal format, and plays that re-encoding. What you hear is an approximation of the original model. How close it lands depends entirely on how good the conversion is.

Native. The device runs the NAM model itself. Nothing is re-encoded. What you hear is the model — the same one the plugin on your computer would run.

Native was historically impractical on affordable pedal hardware, because the original NAM architecture was designed for computers. A2 changed that. NUX MG hardware now runs the native path.

CONVERTED .nam filefull precision Re-encodeproprietary format Approximation playsaccuracy depends on conversion NATIVE — NUX MG .nam filefull precision The NAM model runs on the deviceno conversion step — nothing in between Same model. Fewer steps.

Architecture 2

What changed with A2

A2 is the second generation of Neural Amp Modeler, built by TONE3000 in partnership with Steve Atkinson, NAM’s creator. It was rebuilt from the ground up with two goals: raise accuracy, and cut the processing cost enough that NAM could finally run on the small chips inside mass-market hardware.

Every A2 file carries two sizes in one download. A2-Full is the maximum-accuracy version for DAWs and devices with processing headroom. A2-Lite is the same model run at a smaller size, designed for processing-constrained hardware like multi-effects pedals. The device picks whichever fits its budget — and A2-Lite is what makes native NAM on commercial pedal hardware possible at all.

50% CPUWhat A2-Lite needs on a 600 MHz ARM Cortex-M7 — a chip that costs about $3 in bulk.
1,000+Participants in TONE3000’s MUSHRA blind listening test, submitting over 100,000 ratings.
0.0033Median error (ESR) after retraining the library to A2 — roughly half the A1 figure of 0.0062.

Figures published by TONE3000 from tests run in March 2026. In their blind listening test A2-Lite scored above the commercial modelers they benchmarked against, including Neural DSP, IK Multimedia ToneX and Line 6 Proxy. Their test code and raw ratings data are open source — read the full A2 guide.

Two engines, your call

Deep Image and NAM A2

Your NUX doesn’t make you pick a side. Deep Image is our own profiling technology; NAM A2 is the open community standard. Both live in the AMP block, and you choose per preset.

They are built for different jobs, and the honest difference comes down to one thing: Deep Image captures multiple layers, NAM captures one state. A Deep Image profile consolidates several layers of simulation into a single amp profile, so the gain structure stays responsive after the capture. A NAM model learns your gear exactly as it was set on the day you recorded it — extraordinarily faithful to that setting, but that setting is what you get.

NUX Deep Image NAM A2
Who makes it NUX proprietary black-box profiling Open source, trained by a global community
Where tones come from Capture your own gear with the NUX IMAGE app, plus NUX factory profiles Tens of thousands of free captures on TONE3000, or train your own
After capture Multi-layer — gain structure stays responsive A snapshot of one setting; tone controls sit around it, not inside it
What it can capture Amps, preamps, overdrive and distortion pedals Amps, cabs, full rigs, pedals and outboard gear
Cabinet Pair it with the IR block Usually baked in — the IR block steps aside
Portability Runs on NUX hardware Open format — plays anywhere NAM is supported

Most players end up using both: Deep Image for the amps they own and want to keep tweaking, NAM for the ones they’ll never get their hands on.

For the record

Three technologies, not two

NUX amp tone comes from three separate engines. They get mixed up constantly, so here is the short version of what each one actually is.

WHITE BOX TS/AC-4K & TSAC-HD Models the circuit itself with maths, stage by stage. Real negative feedback, so knobs interact the way a tube amp’s do. Built-in amp models BLACK BOX Image & Deep Image Our own profiling. Capture your amp or drive pedal with the NUX IMAGE app. Deep Image folds several layers into one profile. Your gear, made portable NEURAL · OPEN SOURCE NAM A2 A neural network trained on a recording of real gear. Not a NUX format — an open one, shared by everyone who uses it. The whole community’s gear

Get started

Four steps to your first NAM tone

  1. Update your firmware

    Head to your model’s support section and download the latest firmware for macOS or Windows. Back up your presets first — a firmware upgrade resets user settings.

  2. Install the matching editor

    GEN2 firmware only works with the GEN2 QuickTone editor, so update both together. The editor is on the same support page as the firmware, and it’s what you’ll use to move NAM files onto the device.

  3. Find your tones on TONE3000

    TONE3000 hosts the community library — amps, cabs, rigs, pedals and outboard gear, free to download. Filter for A2 captures, or capture your own gear if you’d rather model something you already own.

  4. Load a slot and play

    Drop the .nam file into one of your NAM slots in the editor, select it in the AMP block, and set your Gain, Bass, Mid, Treble and Master around it. That’s the whole workflow.

Questions

Good to know

Do I have to pay for NAM tones?

No. The TONE3000 library is free to browse and download, and NAM itself is open source under an MIT licence. You can also train your own captures at no cost.

My IR block switched itself off. Is something broken?

That’s expected behaviour, not a fault. When NAM is enabled the IR block is automatically disabled by default, because a NAM capture usually already includes the cabinet simulation. Running both would stack two cabs. If your capture is an amp head only, you can turn the IR block back on.

What happens to my older A1 files?

A2 is a new architecture rather than a drop-in update, so A1 and A2 are handled differently. The MG-400 GEN2 firmware, for example, lists support for .nam files as both A2-Lite Native and A1 Converted models. Check your specific model’s firmware notes for exactly what it accepts.

A2-Full or A2-Lite — do I need to pick?

No. A single A2 download contains both sizes, and the device or application loads whichever suits its processing budget. On NUX pedal hardware that’s A2-Lite. You don’t download anything different.

Does NAM replace Deep Image?

Not at all. Both engines stay available in the AMP block and you choose per preset. See the comparison above for where each one is the better tool.

Can I capture my own amp?

Yes, and it’s the reason a lot of people get into this. You record your amp’s dry input and wet output, then train a model from that recording. TONE3000 walks you through it. You can also capture with the NUX IMAGE app if you’d rather make a Deep Image profile.

Any amp you want.

Update the firmware, pick a capture, plug in.

Neural Amp Modeler (NAM) is open-source software created by Steve Atkinson. TONE3000 is an independent platform and is not affiliated with NUX or Cherub Technology. All brand and model names mentioned on this page are trademarks of their respective owners, which are in no way associated or affiliated with NUX Effects and Cherub Technology Co., Ltd.