Blog
Latest updates and articles
Something is wrong with Sonnet 4.5
We're seeing an elevated number of failed tests in our coding benchmark for Sonnet 4.5. Sonnet 4 looks normal.
IsItNerfed? Sonnet 4.5 tested!
Sonnet 4.5 benchmark results show 46% failure rate compared to Sonnet 4's 37% on our dataset
New Release (Oct 1, 2025): More Models, UI/UX Improvements
Added Gemini and GPT-4o support, separated AI agents from LLMs, and improved mobile UX
New Release (Sept 27, 2025): Charts, theme, and data export
Improvements to charts with zoom and panning, SMA indicator, CSV export, and a fresh new theme
New Release (Sept 22, 2025): Navbar, UI improvements, roadmap, and contact page
Updates include a new navbar, UI improvements, a roadmap, and a contact us page
AI Nerf: Anthropic's Incident Matches Our Data
Anthropic confirms degraded Claude performance during Aug 29-Sep 4, matching our telemetry data
The AI Nerf Is Real
How we discovered volatile performance patterns in Claude Code through real-time LLM monitoring