Gateway installation
The Gateway ships inside the NeuraFrame Studio™ package. You install the same package, then turn on the gateway block and point it at your provider. There is nothing separate to download.
What you need
| Item | Requirement |
|---|---|
| Platform | 64-bit Linux, ARM64 or x86_64 (AWS Graviton, standard x86_64 servers, and Jetson all work) |
| Python | 3.8 or newer (present on the supported images) |
| Upstream | Your model provider's base URL and API key, exactly as your app uses them today |
The Gateway sits in front of a model; it does not include one. Your provider can be a hosted API (OpenAI, Anthropic, xAI) or a local model server. Unlike the reuse engine's own workflows, the Gateway path needs nothing running locally except NeuraFrame™ itself.
Install the package
Download the package for your architecture from the download page, then run the installer as root. This is the same one-line install as Studio.
tar -xzf neuraframe-studio-*-0.3.0.tar.gz cd neuraframe-studio-* sudo bash install.sh
The installer sets up the neuraframe-studio service and the neuraframe command. On a fresh install the Gateway is idle until you give it an upstream.
Turn on the gateway
Set the gateway's upstream to your provider and choose a profile, then restart. The Gateway listens on http://127.0.0.1:8081 by default.
neuraframe config set gateway.upstream_url https://api.openai.com
neuraframe config set gateway.profile auto
sudo systemctl restart neuraframe-studio
neuraframe status # confirm the service and gateway are up
Then change one line in your app: point its base URL at the gateway instead of your provider. Full provider table and profile details are on Configuration & profiles.
Verify
neuraframe doctor # preflight, service, and license summary
Send one request from your app through the gateway, then send the same request again. The second one is served from memory with no upstream call. If a request cannot reach your provider, the Gateway returns a clear error rather than inventing an answer; see Troubleshooting.
Updating to a new version
The Gateway ships inside the Studio package, so you update it the same way: install the newer package over your current one. Your license, configuration, and reuse memory are preserved, and the memory store migrates itself, so there is nothing to redo. Only neuraframe uninstall --purge removes that state.
tar -xzf neuraframe-studio-*.tar.gz cd neuraframe-studio-* sudo bash install.sh neuraframe status