Kagi LLM Benchmarking Project
Introducing the Kagi LLM Benchmarking Project, which evaluates major large language models (LLMs) on their reasoning, coding, and instruction following capabilities on an unpolluted benchmark relevant to using LLMs in the context of search.
Improvements and bug fixes
- Fixes to session synchronization between devices
- Fixed a performance regression on initial search page load
- Adjust the redirect for "no results found" when searching with Lenses #4471 @snuser
- Comboboxes' search bar is overlapped by provided options below #4472 @snuser
- !gi redirection broken from image result pages #4486 @tuesday
- Non ascii characters breaks from image search #4434 @etra0
- Image search filters not working #4424 @-mbs-
- Ask questions about the page issue is still not resolved #4519 @ntt
- Search control popup is trapped inside when word definition box is present #4482 @sefidel
- Should be able to click on a Map item and see more about it #4055 @jducoeur
- Some of the text for the widgets are not translatable #4183 @Grooty
- When using upload feature it no longer responds to follow up questions #4447 @cakeboss
- Upgraded Mistal Small to Mistral Nemo model in Assistant chat mode
