Use host.docker.internal to reach local Ollama from Mule

A developer documented that MuleSoft's Inference Connector fails with a connection-refused error when pointed at a local Ollama instance via http://localhost:11434/v1, and that switching the openAICompatibleURL to http://host.docker.internal:11434/v1 resolves it. The working configuration runs the qwen2.5:14b model through the OpenAI-compatible connection in Mule 4 Community kernel 4.12.0, returning a 200 status with the model output in the response field.

The flow failed with a connection refused error when the configuration used http://localhost:11434/v1 as the base URL. The MuleSoft Inference Connector enables text generation workflows docs https://docs.mulesoft.com/mulesoft-inference-connector/latest/ . { "prompt": "Reply with exactly one word: pong" } <?xml version="1.0" encoding="UTF-8"? <mule xmlns="http://www.mulesoft.org/schema/mule/core" xmlns:http="http://www.mulesoft.org/schema/mule/http" xmlns:ms-inference="http://www.mulesoft.org/schema/mule/ms-inference" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation=" http://www.mulesoft.org/schema/mule/core http://www.mulesoft.org/schema/mule/core/current/mule.xsd http://www.mulesoft.org/schema/mule/http http://www.mulesoft.org/schema/mule/http/current/mule-http.xsd http://www.mulesoft.org/schema/mule/ms-inference http://www.mulesoft.org/schema/mule/ms-inference/current/mule-ms-inference.xsd" <http:listener-config name="http-listener-config" <http:listener-connection host="0.0.0.0" port="8081" / </http:listener-config <ms-inference:text-generation-config name="ollama-openai-compatible" <ms-inference:openai-compatible-connection openAICompatibleModelName="qwen2.5:14b" openAICompatibleURL="http://host.docker.internal:11434/v1" apiKey="ollama" maxTokens="16" temperature="0" topP="1" timeout="120" / </ms-inference:text-generation-config <flow name="ask" <http:listener config-ref="http-listener-config" path="/ask" / <ms-inference:chat-answer-prompt config-ref="ollama-openai-compatible" <ms-inference:prompt payload.prompt </ms-inference:prompt </ms-inference:chat-answer-prompt </flow </mule {"response": {"response": "Pong"}, "status": 200} The response field contains the model output, and status shows 200. Using http://localhost:11434/v1 as the base URL causes a connection refused error. I configure openAICompatibleURL to http://host.docker.internal:11434/v1 in the connector configuration. Runs shown: Mule 4 Community kernel 4.12.0 104 DataWeave patterns + 8 Exchange modules with 208 MUnit tests: github.com/shakarbisetty/mulesoft-cookbook | 60-second walkthroughs: youtube.com/@SanThaParv