Changelog

A running record of what's new in Keimodel — features, fixes, and improvements.

v0.5.0May 29, 2026
  • Added animated hero text on the landing page for a more engaging first impression
  • Analysis style picker now remembers your last selection across sessions
  • Model selector now highlights the active analysis style when you re-open settings
  • New provider filter in the model picker so you can browse by company
  • Long model names are now fully readable — no more truncation in tight layouts
v0.4.0May 15, 2026
  • Response streaming: model replies now appear word-by-word as they arrive
  • Tokens-per-second metric added to each panel so you can compare model efficiency
  • Fastest model is now highlighted with a subtle badge on the response card
  • Response time is included in the analysis so the AI can factor in speed
  • Download all responses together as one formatted file
v0.3.0April 28, 2026
  • History panel: past comparisons are saved and can be reloaded at any time
  • Pin and archive history entries to keep your workspace organised
  • Users can now pick their own default models from the settings page
  • Added social proof section to the landing page
  • Credits modal can now be reopened directly from the low-balance notice
v0.2.0April 10, 2026
  • Launched batch comparison mode for running the same prompt across many models
  • Added code-analysis style with correctness, quality, and security breakdowns
  • Custom system prompt support added to the comparison panel
  • Dark / light theme toggle with preference persistence
  • Introduced the credit system with Starter, Pro, and Max bundles
v0.1.0March 25, 2026
  • Initial launch of Keimodel — compare responses from multiple LLMs side-by-side
  • Support for OpenAI, Anthropic, Google, Meta, Mistral, DeepSeek, xAI, and more
  • AI-powered analysis that highlights consensus, conflicts, and key insights
  • Clerk authentication with Google and GitHub sign-in
  • Responsive layout with up to 6 model panels simultaneously