DevBridge AI Provider for Magnitude

Magnitude lists only models that are downloaded and currently running. Open the Magnitude application, download a model if necessary, start it, and reload the Magnitude settings page.

No. When Magnitude runs on the same computer as WordPress, leave the API key field under Settings > Connectors empty. The field may be displayed as part of the standard connector interface, but Magnitude does not require a key for local use.

1. Open the Magnitude application on the computer running the model.
2. Enable Network access in Magnitude.
3. If the Require API key option is enabled, copy the API key and enter it in the Magnitude connector under Settings > Connectors in WordPress. Save your settings.
4. Find the network address shown under Reachable at in Magnitude. The address should end in /inference/v1.
5. Open Settings > Magnitude in WordPress and enter the reachable address. The plugin handles the /v1 API path, so either the base /inference address or the /inference/v1 address can be used.

Reasoning models may use a significant number of tokens for internal reasoning before producing their final answer. Increase the maximum token limit for the feature you are using.

No. Magnitude provides text and chat inference through this connector, not image generation. If you need image generation or embeddings in WordPress, consider an AI provider that supports those capabilities, such as a compatible Ollama setup.

Yes, when the running model supports the relevant input type and Magnitude reports that capability. Support depends on the model and the capabilities exposed by Magnitude.

Both Magnitude and Ollama can run open-weight models locally and provide OpenAI-compatible APIs. They are separate applications, and each has its own WordPress connector.

• Focus: Magnitude is an inference engine designed to optimize inference for the available hardware. Ollama offers a broad library of models and also provides a cloud service.
• Performance: Magnitude publishes benchmarks showing performance improvements over llama.cpp on tested hardware. Actual performance depends on your hardware, model, and configuration.
• Default address: This plugin uses http://127.0.0.1:10100/inference. The Ollama API commonly uses http://localhost:11434.
• WordPress features: This plugin provides text generation, tool calling, structured JSON output, and supported image and audio input. Image generation and embeddings are not available through Magnitude via this connector.

Use the connector that matches the inference application you run. Choose this plugin if you use Magnitude or want to try it. If you need image generation or embeddings, choose a provider that supports those features, such as the Ollama connector.

Conversation history: When you test models in the Test Playground under Settings > Magnitude, the multi-turn conversation history is stored in your browser's session storage (sessionStorage). It is not stored in the WordPress database. Clearing the history or closing the browser tab removes the session history. Response cache: To reduce repeated inference requests while testing, responses from the Test Playground are cached using WordPress transients for up to 10 minutes. Enable Bypass Cache to skip cached responses and request a fresh result. Uninstall cleanup: When the plugin is deleted, its saved options and temporary transients are removed from the WordPress database on single-site and multisite installations.

No. DevBridge AI Provider for Magnitude is an independent project. It was inspired by the AI Provider for Ollama plugin but is not affiliated with Ollama, its makers, or the authors of that plugin.