Jev System One Model vs LLMs: Benchmarks and When to Use It
Jev, TypeSafe AI's System One Model, matches Sonnet 5's 67.8% on workflow evals at $0.0004 vs $0.1174 per case. How it works and when to use it.
6 posts
Notes on tech, personal growth, life experiences, and the occasional reflection. I write about things I'm learning, hoping that it helps someone else on a similar path.
Jev, TypeSafe AI's System One Model, matches Sonnet 5's 67.8% on workflow evals at $0.0004 vs $0.1174 per case. How it works and when to use it.
A practical guide to AI agent skills and the Model Context Protocol. What each one is, what it costs you in context, and how to pick the right one without overthinking it.
A look back at the successes, lessons, and hurdles of 2024, along with a roadmap for personal and professional growth in 2025.
Reflection on the accomplishments and challenges of the year 2023 and outline the goals for the next year.
Handle promise errors in TypeScript: then/catch, try/catch, a typed result tuple, allSettled, AbortSignal timeouts and retries, with real output.
Add dark mode to an Astro (or any) site with Tailwind CSS v4 and CSS variables, no `dark:` prefixes. Follows prefers-color-scheme, with a toggle.