How to use Darkbloom with n8n

At a glance

How do I use Darkbloom with n8n?

n8n has no Darkbloom node (a search for it in n8n 2.44.6 finds nothing). It does not need one: n8n’s OpenAI credential has a Base URL field, and Darkbloom answers the same requests as OpenAI’s API. So you point an OpenAI credential at Darkbloom and use n8n’s OpenAI nodes. The steps below describe n8n 2.44.6, which we ran in Docker on our own machine. Menu names can differ in other versions, and we did not run n8n Cloud.

1. Get a Darkbloom API key. Open the Darkbloom console and sign in, add credits (the minimum is $0.50, and Darkbloom says credits are non-refundable and never expire), then choose API console in the console’s left menu and Create API key (Darkbloom’s docs call this page Settings → API Keys). The console shows the key only once, so copy it right away. Every key spends from your prepaid balance, so add only what you are willing to spend while you test, and watch your first runs, because an AI Agent can send several requests for one message.
2. Create an OpenAI credential that points at Darkbloom. In n8n, click + at the top of the left bar and choose New credential. Type OpenAI, choose OpenAI and click Continue. Click the name at the top, OpenAI account, and rename it Darkbloom, so you can tell it from a real OpenAI credential. Paste your key into API Key, and replace the address in Base URL with the one below. Leave Organization ID empty and everything else as it is, then click Save.

   API KeyCopy

   YOUR_DARKBLOOM_API_KEY

   Base URLCopy

   https://api.darkbloom.dev/v1

3. Read the connection test. When you save, n8n tests the credential by asking the Base URL for its model list. Connection tested successfully means Darkbloom accepted the key. Couldn’t connect with these settings with Unauthorized means the key is wrong, and with Not Found it means the Base URL does not end at /v1. n8n saves the credential either way, so fix the field and save again.
4. Build a small workflow. Click + at the top of the left bar and choose New workflow. Click the large + box, choose On chat message, and close its panel. Click the + after that node, search for AI Agent and choose it. At the bottom of the AI Agent panel, click the + under Chat Model and choose OpenAI Chat Model.
5. Choose the model. In the OpenAI Chat Model panel, check that Credential shows Darkbloom. n8n fills it in when it is your only OpenAI credential. If you choose another one later, n8n empties Model, so choose the credential first. Then open Model, keep From list and choose a Darkbloom model. n8n fills the list from Darkbloom, sorted by name, for example gemma-4-26b. Do not leave the preset, gpt-5-mini: it is not a Darkbloom model.

   ModelCopy

   gemma-4-26b

6. Turn off Use Responses API. A new OpenAI Chat Model node in n8n 2.44.6 has this switch on, so it sends its requests to Darkbloom’s /v1/responses. That also worked in our checks, but Darkbloom’s docs describe only chat completions, so we switch it off to use /v1/chat/completions. Switching it off also hides Built-in Tools. We tried one of them, Web Search, and Darkbloom answered 400.
7. Send a message. Close the panel. The chat is at the bottom of the canvas, with the Logs panel next to it. If the chat is hidden, click Open chat at the bottom of the canvas. Type Say hello in five words. in the message box and press Enter. The reply appears in the chat, and Logs lists the nodes that ran, how long they took and how many tokens they used.
8. Optional: limit the length of the answers. In the OpenAI Chat Model panel, click Add Option under Options and choose Maximum Number of Tokens. Enter the most tokens you want back, with room for thinking if your model thinks first (see the models section).

What does each setting in n8n mean?

The first rows are the fields of the credential and the rest are the node settings that mattered when we tested. We left every setting that is not listed at its default.

Which n8n nodes work with Darkbloom?

We ran the text nodes. Darkbloom’s API has no routes for several of the other things the OpenAI credential offers, as the last rows show.

Which Darkbloom model should I use with n8n?

Use any ID that GET /v1/models returns. On October 23, 2026 it listed 12 models, shown here from the cheapest output price up, in US dollars per million tokens. Check the live prices before you rely on a number; that public list also names models that are not online. A key can be limited to some models, and Darkbloom refuses any other with a 403 (see the errors table).

n8n’s Model list shows every model that Darkbloom returns, so a listed model can still be refused with a 403 if your key is limited to some models. An AI Agent sends tool definitions when you connect a tool, so choose a model whose Features column lists tools. We ran gemma-4-26b for every check, and EigenLabs/Qwen3.8-27B-4bit-mtp for one-line messages through both APIs. A model that thinks first spends part of the token limit on its thinking: with Maximum Number of Tokens at 30, EigenLabs/Qwen3.8-27B-4bit-mtp used all 30 and n8n showed { "output": "" } as a success.

How much does n8n cost with Darkbloom, and how do I keep it low?

Darkbloom bills the tokens you use from your prepaid balance: (input tokens × input price + output tokens × output price) ÷ 1,000,000, with a minimum of $0.0001 per request. n8n adds little to a request beyond your message, the agent’s instructions and the definitions of the tools you connect. A one-line message to gemma-4-26b was 19 tokens in and 7 or 8 out, depending on the reply, which at the October 23, 2026 prices is worth no more than $0.000003, so Darkbloom billed the minimum, $0.0001. An AI Agent with the Calculator tool answered one question with two requests (95 tokens in and 22 out, then 136 in and 27 out), and each was billed the minimum. A model that thinks pays for its thinking: EigenLabs/Qwen3.8-27B-4bit-mtp used 84 tokens out for a five-word reply, and Darkbloom billed $0.000186.

The token count in n8n’s Logs panel is only a guide, and the bill is Darkbloom’s count. With chat completions it matched: the panel showed 26 tokens for a request that Darkbloom logged as 19 in and 7 out, and 27 for 19 in and 8 out. With Use Responses API on, it showed “~14 Tokens”, an estimate.

Check your real spend in the Darkbloom console, or call GET /v1/payments/usage with your key (Darkbloom docs: usage history).

Where does n8n keep my API key?

Encrypted, in n8n’s own database. n8n makes an encryption key the first time it starts and saves it in a config file in its data folder (~/.n8n, which is /home/node/.n8n in Docker). With n8n’s documented Docker command, that file could be read only by its owner (mode 600). In our test the key text did not appear in the database files, and the n8n screen shows it masked. Anyone who can use the credential in n8n can run workflows with your key, so share it only with people who should spend your balance. If the key leaks, delete it in the Darkbloom console and create a new one.

What do the common errors mean?

In the chat panel, a failed run says only Error in workflow. Look at the Logs panel next to the chat and click OpenAI Chat Model: its Output shows n8n’s sentence first and Darkbloom’s own answer on the line below it. n8n also pops up a message that starts with Problem in node and the name of the AI Agent, followed by the first sentence. Each row was reproduced with n8n 2.44.6, except the empty-balance row.

Is Darkbloom private, and is it ready for production?

Not yet for sensitive work. Darkbloom’s quickstart calls the service “an experimental research prototype” and says not to use it in production applications, and its console labels it Public alpha. Darkbloom says prompts are encrypted to a verified provider Mac. Its security overview says the coordinator re-encrypts each request for the chosen provider and runs in a confidential VM. An independent review by Wavect (September 2026) reads Darkbloom’s technical documentation as saying that both the coordinator and the provider handle your prompt as plaintext inside protected boundaries.

n8n sends the text of your message, the earlier messages if you connect a Memory node, the instructions of the AI node, its tool list and each tool’s output to Darkbloom. Do not send secrets, customer data or code you are not allowed to share.

n8n keeps a copy of your test messages and replies for itself. In our test, the text of the messages was readable in n8n’s database files and in its event log, all in the data folder (~/.n8n). The key was not. Protect that folder and its backups like the prompts themselves.

How this guide was checked

• A paid request from n8n to api.darkbloom.dev worked on October 23, 2026. We ran the steps on this page again as written, from a clean state. n8n 2.44.6 in Docker, with an empty data folder, answered a chat message through an AI Agent with gemma-4-26b chosen from the Model list, again with the ID typed by hand, and again with Maximum Number of Tokens set to 50. Darkbloom’s usage log shows those three requests billed $0.0001 each, the minimum.
• The setup steps were compared with n8n’s own docs (linked under Sources) on October 22, 2026. Where the docs and the program differ, this page follows what we saw: n8n’s OpenAI credential page lists only an API key and an Organization ID, and the form also has Base URL, Add Custom Header and Allowed HTTP Request Domains; the OpenAI Chat Model page says the node defaults to the Chat Completions API, and a new node in 2.44.6 had Use Responses API switched on and gpt-5-mini chosen; the OpenAI node page lists “Generate a Chat Completion” and “Generate a Model Response”, and the node in 2.44.6 offers “Message a model” and “Classify text for violations”; and the AI Agent page says you must connect a tool, while an agent with only a chat model ran.
• We ran n8n 2.44.6, the stable version in n8n’s docs that day, in Docker, with n8n’s documented command and these changes for our machine: n8n’s image on GitHub’s registry (ghcr.io/n8n-io/n8n) because Docker Hub was limiting our test machine, the host network, a listening address without IPv6, our network’s proxy and certificate file, and switches that turn off version notices, diagnostics, templates and the hiring banner. We used a browser from a fresh start: the owner account screen, the credential as on this page, a workflow with a chat trigger, an AI Agent and an OpenAI Chat Model, and chat messages. We also ran a Basic LLM Chain, a Calculator tool, a Simple Memory node, the Maximum Number of Tokens, Timeout and Max Iterations options, streaming, a built-in tool, and the OpenAI node’s Message a model action. To see what n8n sends, we also ran it through a local forwarder that passed each request to Darkbloom unchanged and logged its path, size and settings.
• On October 23, 2026 we ran the steps again, in the order of this page, in a new container with an empty data folder: the credential from the + menu (its test said Connection tested successfully), the workflow from the + menu, the model chosen from the list and typed by ID, Use Responses API off, and Maximum Number of Tokens at 50. Left at its preset, gpt-5-mini, the same workflow gave the 403 of the first row in the errors table. Darkbloom’s usage log shows the three requests that it accepted, each billed $0.0001, and no row for the refused one.
• The error rows were reproduced with n8n against Darkbloom, except the “invalid API key” rows, which used a test server that returns Darkbloom’s real reply for a wrong key, and the empty-balance row, because that needs an empty balance. Darkbloom’s 404 for /v1/embeddings, /v1/moderations, /v1/files and /v1/conversations comes from calling it directly, not from n8n.
• The request sizes and amounts in the cost section come from Darkbloom’s usage log. We matched each request to its row by the job ID that Darkbloom returns. On the key storage, we looked in n8n’s database files, its event log and the container’s log for the key text and for a test message, and exported the workflows and credentials with n8n’s command line.
• That Darkbloom lists POST /v1/responses comes from Darkbloom’s console, which names it in its API console page, and from the README of its source code; its docs site documents only chat completions. We called it with n8n and it worked for the two models we tried.
• The model list, context windows, features and prices come from GET /v1/models on October 22, 2026 (it needs a key). The key steps in the console come from the console’s own page text. We tested with a key that already existed, so we did not click through key creation ourselves.
• Our test machine adds that key to requests to Darkbloom on their way out, so in the runs against Darkbloom the credential held this page’s placeholder. That n8n sends the key it is given was checked separately, with a local test server that prints the Authorization header it receives: it got Bearer and the key we had entered.
• Not run: n8n Cloud, the npm and desktop installs, Windows, publishing a workflow and running it from a trigger other than the chat panel, the Embeddings, Files, Conversations, moderation, image, audio and video actions, a vector store, retries, other n8n versions, and an empty balance.
• Our Darkbloom link was opened in a clean browser profile on October 22, 2026. It saved the referral code and the page said the code “is saved and will be applied after you sign in”. We did not create an account to test the sign-in step itself.

About this guide

BloomGauge is a free Mac app for people who run Darkbloom providers, so we benefit when Darkbloom has more paying demand. We are independent of Darkbloom and not endorsed by it. Our Darkbloom sign-in links carry our referral code, BLOOMGAUGE. If you sign in to Darkbloom after opening one, Darkbloom attaches the code to your account and pays us a share of your eligible token spend. Darkbloom funds that reward, so it does not change your price.

Sources

• n8n docs: OpenAI credentials
• n8n docs: OpenAI Chat Model node
• n8n docs: OpenAI node
• n8n docs: AI Agent node
• n8n docs: install with Docker
• Darkbloom console: API console (endpoint list)
• Darkbloom source code: README (API surface)
• Darkbloom docs: quickstart
• Darkbloom docs: chat completions
• Darkbloom docs: authentication and API keys
• Darkbloom docs: add credits
• Darkbloom docs: usage history
• Darkbloom live prices (GET /v1/pricing)

Found something out of date? Tell us in #bloomgauge on the Darkbloom Slack or at Support.