Works with the tools you already use.

Every integration on this page comes down to the same three values. Where a tool asks for an "OpenAI-compatible" or "custom OpenAI" provider, these are what go in the fields.

The OpenAI SDKs take a base URL in their constructor. Nothing else changes: streaming, tool definitions and image inputs use the same request shapes you already know.

import os
from openai import OpenAI

client = OpenAI(
    base_url="https://neurongate.net/v1",
    api_key=os.environ["NEURONGATE_API_KEY"],
)

stream = client.chat.completions.create(
    model="anthropic/claude-sonnet-5.5",
    messages=[{"role": "user", "content": "Explain idempotency keys in two sentences."}],
    stream=True,
)
for chunk in stream:
    if chunk.choices and chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="")

Non-streaming requests can carry an Idempotency-Key header. A retry with the same key and body returns the stored response instead of running (and billing) the request twice.

Frameworks either wrap the OpenAI SDK or have an "OpenAI-compatible" class. Use that class and pass the base URL; the model string is the NeuronGate id.

# pip install langchain-openai
import os
from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
    model="anthropic/claude-sonnet-5.5",
    base_url="https://neurongate.net/v1",
    api_key=os.environ["NEURONGATE_API_KEY"],
)
print(llm.invoke("Summarise the CAP theorem.").content)

Editors that let you bring your own OpenAI key can use NeuronGate by overriding the base URL. Menu names move between releases; the fields below are what to look for.

Cursor

1. 1

   Open the models settings

   In recent versions: Settings → Models, then the API keys section.
2. 2

   Enter your key and override the base URL

   Paste your sk-ng-… key into the OpenAI API key field, enable the option to override the OpenAI base URL, and set it to https://neurongate.net/v1.
3. 3

   Add the model ids you want

   Add a custom model named exactly as in the catalog, e.g. anthropic/claude-sonnet-5.5, and select it in chat.

Continue

Continue reads models from config.yaml. Use the openai provider with an apiBase.

name: NeuronGate
version: 0.0.1
schema: v1
models:
  - name: Claude Sonnet (NeuronGate)
    provider: openai
    model: anthropic/claude-sonnet-5.5
    apiBase: https://neurongate.net/v1
    apiKey: ${{ secrets.NEURONGATE_API_KEY }}
    roles:
      - chat
      - edit
      - apply

Self-hosted chat front ends give a team one place to use every model. Both of these read the model list from /v1/models, so new releases appear without editing config.

Open WebUI

1. 1

   Add an OpenAI connection

   In recent versions: Admin Panel → Settings → Connections, then add an OpenAI API connection.
2. 2

   Enter the URL and key

   URL https://neurongate.net/v1, key sk-ng-…. Save, and the model picker fills with the catalog.

docker run -d -p 3000:8080 \
  -e OPENAI_API_BASE_URL=https://neurongate.net/v1 \
  -e OPENAI_API_KEY=$NEURONGATE_API_KEY \
  -v open-webui:/app/backend/data \
  --name open-webui \
  ghcr.io/open-webui/open-webui:main

LibreChat

Add NeuronGate as a custom endpoint in librechat.yaml. With fetch: true LibreChat loads the full model list; default is what shows before that and if fetching fails.

version: 1.2.1  # keep the config version your LibreChat release expects
endpoints:
  custom:
    - name: "NeuronGate"
      apiKey: "${NEURONGATE_API_KEY}"
      baseURL: "https://neurongate.net/v1"
      models:
        default: ["anthropic/claude-sonnet-5.5", "openai/gpt-6-sol"]
        fetch: true
      titleConvo: true
      titleModel: "current_model"
      modelDisplayLabel: "NeuronGate"

Low-code tools usually have an OpenAI credential or an "OpenAI-compatible" provider with a URL field.

Ids have the form provider/model and are listed with their prices, context windows and capabilities in the model catalog and at GET /v1/models.

Configure the id as a setting rather than in code, so switching models is a config change. Providers release often; see how routing works for what happens when a model is retired.

POST /v1/chat/completions accepts the fields below. Any other field is rejected with a 422 that names it, rather than being silently ignored.

Errors come back as JSON with the HTTP status, a detail message and a request_id you can quote to support. SDKs raise their usual exception for the status code.

{
  "error": {
    "status_code": 402,
    "detail": "Insufficient balance. Please add funds to continue.",
    "request_id": "4f1c2b9e-…"
  }
}