Models

Default Assistant
My Custom Prompt "leaks" into the ai answers.
Language-Agnostic Information Retrieval
Kagi Assistant Model Selection Modes (Basic vs. Advanced)
Extended Documentation/Transparancy _(particularly regarding privacy)_ For the Experimental Research Agent
Malformed Experimental Researcher response?
Assistant reaches limit unusually fast (Research)
Gemini Pro 3 regularly stops at the "thinking step", never giving an answer
GLM model doesn't understand how it receives images
GLM-4.7 goes mad when asked about XML elements
The first letter(s) of Grok 4 responses are cut
Bug Report: Custom Instructions Not Available in Experimental Research Assistant
Assistant's answers sometimes contain single words in the wrong language
Copy/save a specific Kagi Assistant response(s) and save to a tag(s)
Assistant can get stuck thinking with no way to stop
Transparency for Kagi Research models
Some GLM-4.6 (reasoning) responses cut off very quickly
Research Assistant fails yet again to adhere to initial requirements
Assistant tokens information incorrect
Gemini Flash 3 availability