Added provider presets for OpenRouter, Groq, NVIDIA NIM, Cerebras, SambaNova, OpenCode Zen, OpenCode Go, and native Gemini so only the model, API key, and timeout need to be entered manually.
2.2.0
Added Koneek as a standalone WordPress AI Connector with its own provider identity and connector logo.
Removed the hard dependency on the AI plugin while retaining compatibility with its editor features.
Added a Connectors screen summary for all saved Koneek connections with direct access to Koneek settings.
Added Koneek connector credentials and preferred-model integration for reliable AI feature detection.
Improved model discovery and fallback routing across OpenAI, Anthropic, Google, and OpenAI-compatible connections.
Added Server-Sent Events (SSE) response support for OpenAI-compatible AI gateways such as OmniRoute and 9Router, while keeping reasoning content separate from the final answer.
Fixed AI plugin settings incorrectly reporting that no valid AI Connector was configured when Koneek already had a saved API key.
2.1.0
Fix Koneek that still relying on the separate OpenAI provider plugin.
Added model preference integration so the WordPress AI plugin can detect Koneek-configured text generation models.
Improved OpenAI-compatible request routing and response parsing for Responses API and Chat Completions formats.
2.0.0
Added support to save multiple API Keys with priority order.
Redesigned the settings UI into compact, independent provider cards.
Collapsed connections now display as a compact tile grid showing the name and model, so many providers fit without long stacked cards.
An expanded connection spans the full width for editing; collapsed tiles flow side by side.
Added an optional custom name for each connection, shown in the card header.
Cards are now collapsible. Saved connections start collapsed, showing just the name and priority, so the list stays compact.
1.1.0
Added support for multiple provider connections, each with its own model and priority.
Added priority-based automatic fallback: requests try priority 1 first and fall back to the next connection on timeout or error.
Added cross-provider fallback via an automatic OpenAI/Anthropic/Google request and response translation layer.
