GoRunner docs
Search or askSearch⌘K

Ollama

Message models on your own Ollama server or Ollama's cloud, ask about images, create embeddings, and manage the server's models.

9 actions · Paste an API token

Connect Ollama

  1. In the sidebar, open Connections and press Add connection.
  2. Type ollama and choose Ollama.
  3. Fill in the fields. Each one says Required or Optional.
  4. Press Test and save. GoRunner checks the connection before it saves anything.

The sheet asks for:

Server URL · Required

Where Ollama answers, such as http://ollama.example.com:11434 or https://ollama.com. The server must be reachable from the internet: localhost and private addresses (10.x, 192.168.x, …) are refused unless the installation allows private networks.

API Key · Optional

Needed for ollama.com, or for a server behind a proxy that asks for a bearer token. Leave empty for a server without one.

Enter the address of your Ollama server, or https://ollama.com with an API key (ollama.com → Settings → Keys) for Ollama’s cloud models.

Connections says how to sign in again, rename or delete it. In a product that embeds GoRunner, Add connection can be missing: then your provider makes connections for you.

Do things in Ollama

Add an Ollama step, choose a Resource, then an Operation. Choose a row to see its fields.

To start a workflow from Ollama, use a Schedule, Webhook or Form trigger.

Later steps read a step’s output by its name: see Use data from earlier steps.

Limits

  • A reply from Ollama can be up to 16 MB (a file download can have its own limit). A larger one fails the step; it is never cut short.
  • When Ollama says to slow down (429), the step waits as long as it asks, if the step’s time allows. Otherwise it fails and says when to try again.

Not in your list? See When an app isn’t here.

For developers

In the API, which calls apps pieces, Ollama is pieceId: "ollama" with actionId: "ollama", set by resource and an operation_… field, such as operation_embedding. Get an app returns every field.

It can change something outside the run (writes_external). See How versions and side effects work.

Ollama › Text

Message a Model

Model · Required

The model that answers, as the server names it (gemma4, llama3.2:3b …).

Prompt · Required

What to ask or tell the model.

System Message · Optional

Standing instructions for the model: its role, tone and rules. Optional.

Earlier Messages · Optional

Earlier turns of the conversation, oldest first, sent before the prompt: [{"role":"user","content":"…"},{"role":"assistant","content":"…"}].

Output Format · Optional

Text, or JSON the step parses for later steps (under json).

Options: Text, JSON, JSON matching a schema.

Default: Text.

JSON Schema · Required

The JSON Schema the answer must match, such as {"type":"object","properties":{"city":{"type":"string"}},"required":["city"]}.

Shows when Output Format is JSON matching a schema.

Thinking · Optional

Whether a thinking model reasons before it answers (its thinking comes back in reasoning). Some models (gpt-oss) take Low, Medium or High instead of On.

Options: Model default, On, Off, Low, Medium, High.

Default: Model default.

Max Tokens · Optional

The most tokens the answer may use (Ollama's num_predict). Empty: no limit beyond the model's.

Temperature · Optional

0–2: lower is more focused and repeatable, higher more varied. Empty uses the model's default.

Top P · Optional

0–1: sample only from the most likely tokens whose probabilities add up to this.

Context Length · Optional

How many tokens of prompt and answer the model holds (num_ctx). A long document needs more than the server's default; more uses more memory.

Stop Sequences · Optional

Text that ends the answer when the model writes it, one per line.

Tools · Optional

Functions the model may ask to call: [{"type":"function","function":{"name":"get_weather","description":"…","parameters":{…}}}]. The calls it asks for are in tool_calls; this step does not call them.

Keep Loaded For · Optional

How long the model stays in memory after the answer: 30s, 10m, 1h; 0 unloads it at once, -1 keeps it. Empty uses the server's setting.

More Options · Optional

Other model settings Ollama documents, as JSON (such as {"seed": 7, "repeat_penalty": 1.1, "num_gpu": 1}).

Ollama › Image

Analyze Image

Model · Required

A model that reads images (one whose details list vision).

Question · Required

What to ask about the images.

System Message · Optional

Standing instructions for the model: its role, tone and rules. Optional.

Earlier Messages · Optional

Earlier turns of the conversation, oldest first, sent before the prompt: [{"role":"user","content":"…"},{"role":"assistant","content":"…"}].

Image URLs · Optional

Links to the images, one per line (https:// or data: URLs). The step downloads each link and sends the picture to Ollama.

Image File · Optional

Upload a file, or switch to an expression to use one from an earlier step (such as {{steps.download.result}}). JPEG, PNG, GIF or WebP. Use this, Image URLs, or both.

Output Format · Optional

Text, or JSON the step parses for later steps (under json).

Options: Text, JSON, JSON matching a schema.

Default: Text.

JSON Schema · Required

The JSON Schema the answer must match, such as {"type":"object","properties":{"city":{"type":"string"}},"required":["city"]}.

Shows when Output Format is JSON matching a schema.

Thinking · Optional

Whether a thinking model reasons before it answers (its thinking comes back in reasoning). Some models (gpt-oss) take Low, Medium or High instead of On.

Options: Model default, On, Off, Low, Medium, High.

Default: Model default.

Max Tokens · Optional

The most tokens the answer may use (Ollama's num_predict). Empty: no limit beyond the model's.

Temperature · Optional

0–2: lower is more focused and repeatable, higher more varied. Empty uses the model's default.

Top P · Optional

0–1: sample only from the most likely tokens whose probabilities add up to this.

Context Length · Optional

How many tokens of prompt and answer the model holds (num_ctx). A long document needs more than the server's default; more uses more memory.

Stop Sequences · Optional

Text that ends the answer when the model writes it, one per line.

Keep Loaded For · Optional

How long the model stays in memory after the answer: 30s, 10m, 1h; 0 unloads it at once, -1 keeps it. Empty uses the server's setting.

More Options · Optional

Other model settings Ollama documents, as JSON (such as {"seed": 7, "repeat_penalty": 1.1, "num_gpu": 1}).

Ollama › Embedding

Create Embeddings

Model · Required

An embedding model on the server (embeddinggemma, nomic-embed-text, …).

Texts · Required

The text to embed, or a list of texts from an earlier step (one vector per text, at most 1000).

Dimensions · Optional

How many numbers each vector has, for models that can shorten theirs. Empty uses the model's size.

Cut Long Texts · Optional

Cut a text longer than the model's context to fit. Off: such a text fails the step.

On by default.

Ollama › Model

Get Many and Get Running

Return All · Optional

Read every model, page by page (at most 1000; past that the step fails rather than stop early).

Off by default.

Limit · Optional

How many models to return (1–1000).

Default: 100.

Shows when Return All is off.

Get Details and Delete

Model · Required

The model, as the server names it. To pull, type its name in the Ollama library (gemma4, qwen3:8b, …).

Pull

Model · Required

The model, as the server names it. To pull, type its name in the Ollama library (gemma4, qwen3:8b, …).

Allow Insecure Registry · Optional

Pull from a registry without a valid TLS certificate. Only for a registry of your own.

Off by default.

Copy

Source Model · Required

The model to copy.

New Name · Required

The name the copy gets (such as gemma4-backup).