Use ApexApi in n8n
The ApexApi community node lets any n8n workflow call 129+ models — chat, image, video and audio — across 15+ providers with a single ak- key and one bill. It is a thin client over our OpenAI-compatible API, so the model dropdown stays in sync with the live catalog automatically.
1. Install the node
In n8n, open Settings → Community Nodes → Install, then enter the package name:
n8n-nodes-apexapiWorks on self-hosted n8n and n8n Cloud. Package on npm; n8n's community-node install guide. To update later: Settings → Community Nodes → the installed node → Update.
2. Add your ApexApi credential
Create an API key in your dashboard (it starts with ak-). In any ApexApi node, create an ApexApi account credential and paste the key — it turns green when connected.
3. What you can do
The node exposes four resources:
- Chat → Message a Model — prompt + options: System Prompt, Response Style (Precise / Balanced / Creative), and Output Format (Text / JSON / JSON Custom Schema for clean structured output).
- Image → Generate Image — prompt, size, optional reference image, optional download-to-binary.
- Video → Generate Video — prompt, model-aware duration/resolution, post-and-poll job handling.
- Audio → Generate Speech — text + voice (ElevenLabs).
A separate ApexApi Context node gives your workflows live web context on the same key:
- Read a Page — any URL → clean markdown (auto anti-bot).
- Read a Site — crawl a whole site into markdown (async).
- Structured Extract — ready-made JSON from Amazon, Walmart, Crunchbase, G2, Google Maps, or Zillow (async).
The model dropdown loads live from /v1/models — every model appears the moment it's in the catalog, no node update needed. For agents, the separate ApexApi Chat Model sub-node plugs into n8n's AI Agent "Chat Model" slot.
4. Starter templates
Download a ready-made workflow, then in n8n use Workflows → Import from File and select the .json. Set your ApexApi credential on each ApexApi node, save once, then run.
Scrape any URL into clean markdown, then an AI model summarizes it. The simplest 'web + AI on one key' demo.
Read the guide →Turn an Amazon product URL into structured JSON — title, price, rating, reviews. Swap the scraper for Walmart; map fields into Sheets or a DB.
Read the guide →Crawl a whole website into markdown, then an AI describes what it is — the 'turn a site into an AI knowledge base' pattern.
Read the guide →Pull a product's live price + rating and get a one-line AI verdict. Add a Schedule trigger + Slack alert to make it a monitor.
Read the guide →Crunchbase URL → structured company data → a 2-sentence AI sales brief. Wire your CRM's 'new lead' trigger in front of it.
Read the guide →Grab the latest RSS post, read the full article, then AI summarizes it and drafts a social caption.
Read the guide →Topic → ApexApi writes a LinkedIn post → publishes it. Ships in-canvas setup instructions; swap the Manual Trigger for a Schedule Trigger to post on autopilot.
One trigger fans out to three Chat nodes (Text / JSON / JSON Custom Schema) on the same prompt — run once and compare the outputs side by side.
Reads an RSS feed, writes a caption with ApexApi, and posts it to social via Blotato.
Generates an image with ApexApi and posts it to social via Blotato.
An n8n AI Agent backed by ApexApi as its chat model — switch the underlying model freely.
Templates that post to social use Blotato (one API → X, LinkedIn, Instagram, TikTok, and more); each ships an in-canvas setup note.
For AI assistants (LLM-ready reference)
Paste this into your AI assistant to have it build ApexApi n8n workflows for you.
ApexApi n8n community node
- Install: n8n Settings -> Community Nodes -> Install -> "n8n-nodes-apexapi"
- npm: https://www.npmjs.com/package/n8n-nodes-apexapi
- Credential: "ApexApi account" = API key (starts with "ak-"), from https://apexapi.dev/keys
- Under the hood: OpenAI-compatible API at https://api.apexapi.dev/v1
Node types:
- "n8n-nodes-apexapi.apexApi" (main) — resources: chat | image | video | audio
- "n8n-nodes-apexapi.apexApiChatModel" — AI Agent "Chat Model" sub-node
Models:
- Address as provider/model, e.g. openai/gpt-4o, anthropic/claude-opus-4.8,
google/gemini-3.5-flash, deepseek/deepseek-v4-pro
- Full live list: GET https://api.apexapi.dev/v1/models (no auth needed)
Chat options:
- systemPrompt: string
- responseStyle: precise | balanced | creative
- outputFormat: text | json_object | json_schema (json_schema needs a JSON Schema)
-> maps to OpenAI response_format; works on OpenAI + OpenAI-compatible models
Tips:
- Reasoning models (e.g. openai/gpt-5, *thinking) need a generous token budget.
- After importing a workflow, set the credential and SAVE once before the first run.Troubleshooting
- Empty response on a reasoning model — models like
openai/gpt-5or any "thinking" model spend tokens reasoning; give them a generous token budget so content isn't cut off. - "Cannot read properties of null" on first run — a freshly imported workflow needs a Save once before the first manual run; it runs clean after.
- Can't find the ApexApi Chat Model — it only appears inside an AI Agent node's "Chat Model" slot, not in the normal node search.
- 401 / model not found — check the credential key and that the model ID matches a catalog ID exactly (see the models page).