"Anthropic 'Message a model' node produces no output and sends no network request when executed — no error shown anywhere (UI, console, or Anthropic account activity)."

Technical issue: “Message a model” node (Anthropic) does not execute in n8n Cloud

Context:
I’m using n8n Cloud (hosted cloud version, recent account, 14-day trial period). I have a simple workflow with two nodes:

  1. A Webhook node (trigger), which works perfectly — I’ve successfully tested it via Tally.so, and I can see the JSON data arriving correctly in its output.
  2. An Anthropic → “Message a model” node connected after the Webhook.

Node configuration (Anthropic):

  • Credential: “My own credential” (personal Anthropic API key, created on console.anthropic.com)
  • The Anthropic account has $5 of available credit, no blocked payment card, no visible restrictions
  • Model: claude-sonnet-5, selected via “From list”
  • Prompt: simple test text (Reply only with “Test successful”)
  • Role: User

Exact symptom:
When I click the “Execute step” button (within the node’s panel), nothing happens:

  • No error message appears, either in the n8n interface or in the browser’s developer console (Chrome DevTools)
  • The “OUTPUT” area stays stuck on “No output data”
  • No persistent loading indicator — the button seems to return to its initial state without displaying anything
  • In the Chrome DevTools “Network” tab, no network request to the Anthropic API appears after clicking — suggesting the request isn’t even being sent
  • On the Anthropic account (console.anthropic.com), the “Spending” and “Activity” sections show $0.00 and “No activity in the last 7 days” — confirming that no request has ever reached the API

What I’ve already tried without success:

  1. Tested with n8n’s free credits (“n8n credits”) — same result (nothing happens)
  2. Switched to my own Anthropic API key — same result
  3. Tested in Chrome incognito mode (to rule out extensions/ad blockers) — same result, although in this mode I was initially logged out of my n8n account (401 error on /rest/login), reconnected afterward with no change
  4. Deleted and completely recreated the “Message a model” node from scratch — same result
  5. Verified that the “Model” field is properly filled in (not empty) before each test
  6. Used “Execute workflow” (the general button) in addition to “Execute step” (the node-specific button) — still need to fully test this option

Question:
Why does clicking “Execute step” apparently trigger no outgoing HTTP request to the Anthropic API, with no error message in either the n8n UI or the browser console? Is this a known bug with n8n Cloud’s native Anthropic node, a specific configuration issue, a limitation of the free trial account (14 days), or something else?

Environment details:

  • n8n Cloud, free trial (14 days remaining at time of writing)
  • Browser: Google Chrome (macOS, MacBook Pro)
  • Workflow URL: investiraupays.app.n8n.cloud

Additional detail: I also noticed the “Model” dropdown (“From list”) sometimes

displays an invalid/non-existent model name (e.g., “claude-sonnet-4-6”, which

does not exist in Anthropic’s lineup), suggesting the model list feature might

be pulling incorrect or corrupted data.

Hi @PSO Welcome!
Execute step re-runs the chain from the trigger rather than reusing the Webhook’s earlier output, so the run stops at the Webhook waiting for a fresh test call and the Anthropic node is never reached. No request is ever built, which is why nothing shows up in your Anthropic activity and no error appears anywhere.
Open the Webhook node, send one call from Tally so its output loads, then select Pin data in the OUTPUT panel. With the trigger pinned, Execute step on the Anthropic node runs against that saved data instead of waiting on the webhook.
On Cloud the call to api.anthropic.com goes out from n8n’s servers rather than your browser, so the Network tab was never going to show it either.

Hi @PSO, quick follow up: has this been resolved in the meantime? If not, could you export your workflow (or share a screenshot of the node settings) so we can take a closer look?