---
schema_version: "1"
id: 7e9dc5bdab412ba3e712e21b0b83dd67e8480d9ff81561d0b7d144d7d70fc808
title: MuleSoft Inference Connector Release Notes
collection: mulesoft
version: current
locale: en-us
portal_url: https://docs.salesforce.com/mulesoft/current/en-us/release-notes/connector/mulesoft-inference-connector-release-notes-mule-4.md
canonical_url: https://docs.mulesoft.com/release-notes/connector/mulesoft-inference-connector-release-notes-mule-4
source_site: docs.mulesoft.com
content_type: documentation
content_hash: bf872287205cba9e37c1866b2210f10eea384f676cdef24426db8977c54331f3
description: MuleSoft Documentation Site
latest: true
---
---
product: Release Notes
version: master
---

# MuleSoft Inference Connector Release Notes

> For the full documentation index, see: https://docs.mulesoft.com/llms.txt

Anypoint Connector for MuleSoft Inference (MuleSoft Inference Connector) provides access to
inference offerings for large language models (LLMs) from multiple providers including OpenAI,
OpenAI Compatible Endpoints, OpenRouter, Heroku AI, Azure AI Foundry, Azure OpenAI, and many others.
This connector provides operations to interface directly with the API of various inference
providers, enabling seamless integration of AI capabilities into your Mule applications.

Guide: [MuleSoft Inference Connector User Guide](https://docs.mulesoft.com/mulesoft-inference-connector/latest/)

## 1.2.1

**October 6, 2026**

### Compatibility

<table><colgroup><col> <col></colgroup><thead><tr><th>Software</th><th>Version</th></tr></thead><tbody><tr><td><p>Mule</p></td><td><p>4.9.6 and later</p></td></tr><tr><td><p>OpenJDK</p></td><td><p>17</p></td></tr></tbody></table>

### Fixed Issues

<table><colgroup><col> <col></colgroup><thead><tr><th>Issue Resolution</th><th>ID</th></tr></thead><tbody><tr><td><div><div><p>The <strong>[Image] Generate (only Base64)</strong> operation for OpenAI no longer sends the <code>response_format</code> parameter to GPT image models, which reject it. The operation now sends this parameter only for DALL-E models.</p></div></div></td><td><div><div><p>W-23827390</p></div></div></td></tr><tr><td><div><div><p>These libraries are upgraded to address reported security vulnerabilities:</p></div><div><ul><li><p>The <code>jackson-datatype-jsr310</code> library is upgraded to version 2.22.3.</p></li><li><p>The <code>commons-io</code> library is upgraded to version 2.22.0.</p></li><li><p>The <code>commons-lang3</code> library is upgraded to version 3.21.0.</p></li></ul></div></div></td><td><div><div><p>W-3300013</p></div></div></td></tr></tbody></table>

## 1.2.0

**September 29, 2025**

### What’s New

- The new **\[MCP\] Tooling** operation enables the LLM to select and run tools based on prompts
  without requiring manual inputs.

- The connector now supports the **Swisscom** connection type (beta) in **Text Generation Config**.

- The new **Additional Request Attributes** field allows you to pass a JSON object with additional
  request attributes for all operations. These attributes flatten into the root level of the request
  payload, along with existing standard attributes.

- All references to the NVIDIA company name now comply with the official NVIDIA brand name.

For more information, see
[MuleSoft Inference Connector Reference](https://docs.mulesoft.com/mulesoft-inference-connector/latest/mulesoft-inference-connector-reference).

### Compatibility

<table><colgroup><col> <col></colgroup><thead><tr><th>Software</th><th>Version</th></tr></thead><tbody><tr><td><p>Mule</p></td><td><p>4.9.6 and later</p></td></tr><tr><td><p>OpenJDK</p></td><td><p>17</p></td></tr></tbody></table>

## 1.1.0

**September 8, 2025**

### What’s New

- The connector now supports the **Custom headers** field for all operations. Custom headers always
  take precedence over any existing headers passed as part of an implementation, overwriting values
  where the same key is used.

- The **Nvidia** connection type now includes the custom **\[Nvidia\] Base URL** field for sending
  inference requests, with `https://integrate.api.nvidia.com` as the default. This supports custom
  and self-hosted URLs.

- The **Messages** field in the **\[Chat\] Completions** operation now supports DataSense, which
  provides schema support.

- The **Tools** field in the **\[Tools\] Native Template (Reasoning only)** operation now supports
  DataSense, which provides schema support.

- The **\[Agent\] Define Prompt Template**, **\[Chat\] Answer Prompt**, **\[Chat\] Completions**,
  and **\[Tools\] Native Template (Reasoning only)** operations now provide default support for
  toxicity detection by returning the new `contentFilterResults` and `promptFilterResults`
  additional attributes in the response.

- The connector now supports the **Gemini** connection type (beta) in **Text Generation Config** and
  **Vision Config**.

- In the **Azure OpenAI** connection type:

  - The new **\[Azure OpenAI\] API Version** and **\[Azure OpenAI\] Endpoint** fields are added to
    the **General** tab.

  - In the **General** tab, there is a new section called **Required: Enter URL or Parameters**
    which has the **\[Azure OpenAI\] Endpoint**, **\[Azure OpenAI\] Resource Name**, and **\[Azure
    OpenAI\] Deployment ID** fields. In this section, you can either provide the default or a custom
    endpoint URL (which is used directly and overrides any parameter values), or provide the
    resource name and deployment ID (which are used to construct the base URL).

  - The **\[Azure OpenAI\] User** field is moved from the **General** tab to the **Advanced** tab.

For more information, see
[MuleSoft Inference Connector Reference](https://docs.mulesoft.com/mulesoft-inference-connector/latest/mulesoft-inference-connector-reference).

### Compatibility

<table><colgroup><col> <col></colgroup><thead><tr><th>Software</th><th>Version</th></tr></thead><tbody><tr><td><p>Mule</p></td><td><p>4.9.4 and later</p></td></tr><tr><td><p>OpenJDK</p></td><td><p>17</p></td></tr></tbody></table>

### Fixed Issues

<table><colgroup><col> <col></colgroup><thead><tr><th>Issue Resolution</th><th>ID</th></tr></thead><tbody><tr><td><div><div><p>The <strong>[Tools] Native Template (Reasoning only)</strong> operation now supports a full JSON schema for input parameters, including nested structures and tags such as <code>items</code>, <code>minItems</code>, and <code>uniqueItems</code>.</p></div></div></td><td><div><div><p>W-19271915</p></div></div></td></tr><tr><td><div><div><p>The <strong>[Chat] Completions</strong> operation using the <strong>Vertex AI Express</strong> connection type now supports OpenAI standard input template for chat messages.</p></div></div></td><td><div><div><p>W-19483173</p></div></div></td></tr></tbody></table>

## 1.0.0

**August 1, 2025**

### What’s New

This is a new connector.

For more information, see the
[MuleSoft Inference Connector User Guide](https://docs.mulesoft.com/mulesoft-inference-connector/latest/).

### Compatibility

<table><colgroup><col> <col></colgroup><thead><tr><th>Software</th><th>Version</th></tr></thead><tbody><tr><td><p>Mule</p></td><td><p>4.9.4 and later</p></td></tr><tr><td><p>OpenJDK</p></td><td><p>17</p></td></tr></tbody></table>

## See Also

- [MuleSoft Inference Connector User Guide](https://docs.mulesoft.com/mulesoft-inference-connector/latest/)

- [MuleSoft Inference Connector Reference](https://docs.mulesoft.com/mulesoft-inference-connector/latest/mulesoft-inference-connector-reference)

- [Salesforce Help](https://help.salesforce.com)

- [MuleSoft Services](https://www.mulesoft.com/support-and-services)

- [Known Issues](https://help.salesforce.com/s/issues)