Launching July 27, 2026Get ready →
NexCoder

Download

Download NexCoder

The agentic coding IDE for the Nexa ecosystem - one engine behind six chat modes and a terminal CLI. Available for Windows, macOS, and Linux - pick your platform, verify the checksum, and you are ready to run.

  • Streamed directly from nexcoder.trynexa-ai.com - no third-party redirects.
  • Every installer ships with a published SHA-256 checksum you can verify.
  • Local-first: runs offline against a local model or the hosted endpoint.

Windows

NexCoder for Windows

Download NexCoder for Windows

Loading release details…

Requirements
Windows 10 / 11, 64-bit
Version
…
Installer size
…
Released
…

SHA-256 checksum

…

System requirementsInstallationTroubleshooting

Installation

Three steps from download to first run.

  1. 1

    Download

    Pick your platform above and click Download. Your browser saves the versioned installer - a .exe (Windows), .dmg (macOS), or .AppImage (Linux).

  2. 2

    Verify (recommended)

    Compare the file's SHA-256 with the checksum shown above. Windows: Get-FileHash <file> -Algorithm SHA256. macOS / Linux: shasum -a 256 <file>.

  3. 3

    Install & run

    Windows: run the .exe. macOS: open the .dmg and drag NexCoder to Applications. Linux: chmod +x the .AppImage, then run it.

System requirements

NexCoder targets modern 64-bit Windows, macOS, and Linux.

WindowsWindows 10 or 11, 64-bit (x64)
macOSmacOS 12 Monterey or later (Apple Silicon or Intel)
LinuxUbuntu 20.04+ or an equivalent modern 64-bit distribution
Memory8 GB minimum; 32 GB recommended for the local model server
Disk~1 GB for the app; additional space for local models
GPU (optional)A consumer GPU accelerates the optional local model server
NetworkRequired for the hosted model endpoint; optional when fully local

Troubleshooting

Common questions when installing NexCoder.

First-launch security prompts

Until code signing propagates: on Windows, SmartScreen may warn - choose More info then Run anyway. On macOS, right-click NexCoder and choose Open to clear Gatekeeper. On Linux, mark the .AppImage executable with chmod +x. Always verify the SHA-256 checksum first.

Checksum does not match

Delete the file and download again. A mismatch usually means an interrupted download - the download endpoint supports resuming, so retrying is safe.

Download is unavailable

If the button reports a temporary outage, the release service is briefly unavailable. Wait a moment and retry - your version and checksum reload automatically.

Configuring the model

NexCoder reads its configuration from a .env file beside the executable. Set NEXA_API_KEY there and nowhere else.

Prefer to build from source?

A Python virtual environment plus a frontend build is all it takes. The repository also ships a one-command PyInstaller build.

Run from source

python -m venv venv
venv\Scripts\python.exe -m pip install -r requirements.txt
cd nexcoder\ui && npm install && npm run build && cd ..\..
venv\Scripts\python.exe -m nexcoder.main

Package it yourself

The one-command build writes the packaged app to dist\NexCoder.

python build.py

Models: hosted or local

NexCoder defaults to the hosted, OpenAI-compatible coding-model endpoint. Prefer full offline? Serve a GGUF model locally and point NEXA_API_URL at it - nothing else changes.

Hosted endpoint (default)

Production clients talk to https://nexcoder.trynexa-ai.com/v1 - no local model or GPU required. Set NEXA_API_KEY in your local .env and you are done.

Local GGUF server

The NexCoder MoE coding model (3B active parameters) runs on 32GB RAM plus a consumer GPU by keeping the KV cache in system RAM. The server is OpenAI-compatible, so other coding agents can share it too.

By downloading NexCoder you agree to the license terms and privacy policy.