Skip to main content
View source

Ollama

View as Markdown

A RocketRide LLM node that routes pipeline traffic through a locally-hosted Ollama server.

What it does

Provides text generation against an Ollama server running on your own hardware. The node acts as an llm invoke connection for agents and other nodes that need an LLM, and can also be driven directly via its questions / answers lane pair. Because all inference happens on-premise, no external API key is required, making it a natural fit for privacy-sensitive or air-gapped deployments.

Internally, the node talks to Ollama through its OpenAI-compatible /v1 endpoint using langchain-openai (ChatOpenAI). temperature is configurable (default 0; reasoning models such as gpt-oss auto-use 1.0 when it is left unset), and reasoning_effort can be set for reasoning models. If the configured serverbase URL does not end in /v1, the node appends it automatically, so both http://localhost:11434 and http://localhost:11434/v1 are accepted. The OpenAI client requires a non-empty API key field; the node sends the placeholder string dummy-key, which Ollama ignores.


Configuration

Lanes

Lane inLane outDescription
questionsanswersSend a question directly, receive a generated answer

Fields

Pick a profile from the dropdown; the profile pre-fills model, serverbase, and modelTotalTokens. All three fields are individually overridable when using the custom profile.

FieldTypeDescription
modelstringOllama model
modelTotalTokensnumberTotal Tokens
profilestringDefault "llama3_3". LLM model

Profiles

Llama

ProfileModelContext
Llama 4 Latestllama4:latest10,000,000
Llama 3.3 (default)llama3.3:latest128,000
Llama 3.1 405Bllama3.1:405b128,000
Llama 3.1 70Bllama3.1:70b128,000
Llama 3.1 8Bllama3.1:8b128,000
Llama 3.2 3Bllama3.2128,000
Llama 3.2 1Bllama3.2:1b128,000

Qwen

ProfileModelContext
Qwen 3 Latestqwen3:latest128,000
Qwen 2.5 72Bqwen2.5:72b128,000
Qwen 2.5 32Bqwen2.5:32b128,000
Qwen 2.5 14Bqwen2.5:14b128,000
Qwen 2.5 7Bqwen2.5128,000
Qwen 2.5 3Bqwen2.5:3b128,000
Qwen 2.5 1.5Bqwen2.5:1.5b128,000
Qwen 2.5 0.5Bqwen2.5:0.5b128,000

DeepSeek

ProfileModelContext
DeepSeek R1 671Bdeepseek-r1:671b128,000
DeepSeek R1 32Bdeepseek-r1:32b128,000
DeepSeek R1 14Bdeepseek-r1:14b128,000
DeepSeek R1 7Bdeepseek-r1:7b128,000
DeepSeek R1 1.5Bdeepseek-r1:1.5b128,000

Other

ProfileModelContext
Phi 4 14Bphi416,000
Mistral 7Bmistral32,000

Custom: supply any Ollama model tag, context token count, and server URL directly. The default context for a new custom profile is 16,385 tokens until you change it.


Upstream docs


Schema

FieldTypeDescriptionDefault
modelstringModel
Ollama model
modelTotalTokensnumberTokens
Total Tokens
ollama.profilestringModel
LLM model
"llama3_3"
reasoning_effortstringReasoning Effort
Optional reasoning budget for reasoning models: low, medium, or high. Leave unset to let reasoning models auto-use 'low'; a value set here always wins. Ignored by non-reasoning models.
temperaturenumberTemperature
Sampling temperature. Left unset by default so ollama.py can choose: 0 for standard models, and 1.0 for reasoning models (gpt-oss, deepseek-r1, qwen3, qwq, ...) so they emit a final answer instead of looping on empty output. Set an explicit value here to override the auto behavior.

Dependencies

  • langchain-openai
  • langchain-core
  • langchain