letscodeit.devnotes from the build

HERO

SHORT NOTES

Quick takes, product updates and stray thoughts — the length of a tweet.

NOTES

FOOTER

letscodeit.devnotes from the build

A blog about code, design and the slow art of shipping software that people actually want to use.

PRODUCT

Changelog
Roadmap
API

COMPANY

About
Blog
Notes

RESOURCES

Guest post
RSS feed
Docs

© 2026 letscodeit.dev© 2026 letscodeit.dev — Built in public.

Terms·Privacy·Cookies·License

NOTE

·

Comments

FEATURED POSTS

03 / 03

June 19, 2026

17 min read

We skipped Contentful: GitHub as a CMS for our Next.js blog

We publish letscodeit.dev blog posts from a separate GitHub repository. The Next.js app fetches markdown at runtime with a one-hour cache,…

Dev Team

June 17, 2026

10 min read

GEO for developers: what gets cited by ChatGPT and Perplexity

GEO (generative engine optimization) is how you structure and publish content so AI chatbots and answer engines quote your pages when…

Dev Team

May 28, 2026

9 min read

Token Economics: What Your Text Actually Costs

When you send a message to an LLM, you are not paying for characters or words. You pay for tokens . A token is the unit the model splits…

Dev Team

NOTE

The Kimi K3 model has been shown to be competitive with…

/ notes·/ The Kimi K3 model has been shown to be…

Dev Team·July 21, 2026

The Kimi K3 model has been shown to be competitive with Fable 5, and together, they form a new state-of-the-art approach.

Benchmarking over 1,000 agentic tasks revealed that using a router to combine the strengths of open-source Kimi K3 and closed Fable 5 can achieve a 93% task accuracy rate. K3 excels in terminal, symbolic math, and dev tooling, while Fable is superior in web, data visualization, and multi-language tasks. This combination offers up to 50 times better cost-efficiency compared to using Fable alone.

Source: fireworks.ai

Comments

The Kimi K3 model has been shown to be competitive with… | Let's Code It

Comments (3)

A

Anna Berg·July 21, 2026·edited

Comment deleted

J

Jen L.·July 22, 2026

I've been using Kimi K3 for dev tooling, and it's been a solid performer. Curious how the integration with Fable 5 works in practice.

C

Chris O'Hara·July 22, 2026

The cost-efficiency claim is intriguing, but I'd like to see a detailed breakdown of how they achieve 50 times better savings.

R

Rob H.·July 22, 2026

Anna Berg: Comment deleted

"Benchmarking over 1,000 agentic tasks revealed that using a router to combine the strengths of open-source Kimi K3 and closed Fable 5 can achieve a 93% task accuracy rate." This sounds like it might be too task-specific to generalize.