Inferencer v2.0.6 for MiniMax M3

In version 2.0.6, we added support for MiniMax M3 and GLM 5.2 MTP, as well as Server API image support for OpenClaw and Cherry, drag and drop conversation reordering, and more.

Models

GLM 5.2 MTP is now supported, alongside MiniMax M3. These models bring stronger reasoning, coding and agentic performance, while MTP support helps unlock faster generation where available. We also added support for more DeepSeek V4 community models, giving you a wider range of high-performance local options to choose from.

Server API Image Support

The Server API now supports image input, including integrations with OpenClaw and Cherry. This makes it easier to use Inferencer as a local multimodal backend for agents, chat interfaces and development tools that need to send images alongside text prompts.

Workflow Improvements

You can now drag and drop conversations to reorder them, and queued tasks can also be reordered. This makes it easier to prioritise important chats or jobs when running multiple generations at once. We also improved loop detection and message replay, helping agent-style workflows recover more cleanly when a model repeats itself or a conversation needs to be replayed.

Stability, Caching and Performance

We continued improving engine stability, MTP stability and Distributed Compute stability. We also added caching improvements for Distributed Compute and Kimi K2.7, reducing repeated work across distributed runs and making long-running sessions more consistent, especially when using larger models across multiple machines.

Improvements

• Added support for GLM 5.2, GLM 5.2 MTP and MiniMax M3
• Added Server API image support, including OpenClaw and Cherry
• Added drag and drop conversation reordering
• Added queued task reordering
• Added support for more DeepSeek V4 community models
• Improved loop detection and message replay
• Improved engine and MTP stability
• Improved Distributed Compute stability
• Improved caching for Distributed Compute and Kimi K2.7
• Fixed additional bugs and improved overall performance

As always, if you have any features or suggestions you’re more than welcome to add them to our public roadmap.

Thanks again for your support, and as a reminder, all AI processing is done offline, directly on your devices. No telemetry, no background "update" checks.

P.S. If you find Inferencer useful, please consider leaving a review on the App Store. It would be much appreciated.