A new challenger to rival OpenAI’s best ChatGPT model has arisen from China named DeepSeek R1. The reason it’s causing even more of a stir is because the creators claim DeepSeek R1 was trained for under $5M - a mere fraction of the cost of comparable models to date - and they’ve open sourced the code, the models, all of it.

In the same vein, both TJ and Paige had the chance to try out AI coding assistant Devin.ai firsthand last week. Devin is best described as an energetic junior programmer, and while it offers unique ways of interacting with it: Slack threads, PR comments, and has oversight over multiple repos so it can be asked to do things like compare documentation in one repo to SDK endpoints in another repo, its end value is still questionable.

TypeScript validation libraries have been catching on in recent years, and the creators of some of the most popular ones (Zod, ArkType, and Valibot) have gotten together to promote a common interface for libraries called Standard Schema.

News:

Bonus News:

Fire Starters:

toReversed, toSorted, and toSpliced

What Makes Us Happy this Week:

Paige - The Recruit TV series
Jack - The Subtle Art of Not Giving a F*ck
TJ - Onyx Storm

Thanks as always to our sponsor, the Blue Collar Coder channel on YouTube. You can join us in our Discord channel, explore our website and reach us via email, or Tweet us on X @front_end_fire and BlueSky.

DeepSeek R1, Devin.ai, and TS Validation Standards

Front-End Fire

DeepSeek R1, Devin.ai, and TS Validation Standards