Articles

A quiet corner for essays, system thinking, and work-in-progress notes.

Oct 7, 2026Article

Haiku 5.5 Cut Prices 90%, Then the Fine Print Showed Up

Claude Haiku 5.5 matches GPT-6 Luna at $0.10 per million tokens, but the new tokenizer and a 100K cliff change the real cost. Here is the honest math.

Oct 7, 2026ai

Theo Ported TypeScript to Rust with AI and Never Read the Code

ts-rust is Theo's AI-written Rust port of the TypeScript compiler, built for $24k after a failed $400k attempt. Here is how it benchmarks and what it costs you.

Oct 6, 2026opensource

GPUI by Zed: The Electron Killer That Never Left Home

GPUI powers Zed with 120 FPS GPU rendering and no Electron. Here is why the Rust UI framework stalled, and what it means if you want to build on it.

Oct 6, 2026agents

Dots, Muse, and Grok Bot Shipped in the Same Seven Weeks

OpenAI Dots, Meta Muse, and Grok Bot all landed within weeks. Here is how each always-on agent differs on cost, features, and what users actually said.

Oct 5, 2026ai

Kev and Laya: The Open Source Answer to TypeSafe's Jev

Jev is closed and waitlisted. Kev and Laya are the open source Jev alternatives worth running. Here is how they benchmark, and where they lose.

Oct 4, 2026ai

Jev by TypeSafe AI: the hype, reactions, and two-week clone war

Jev TypeSafe AI's decision model returns typed answers in milliseconds. See the hype, the Reddit skepticism, and the two-week OpenAI and AWS clones.

Oct 4, 2026kimi

Kimi K2.7-Code is Out. The Open-Source Coding Model That Thinks Less

Kimi K2.7-Code is Moonshot AI's latest open-source coding model. We compare it against GPT-5.5 and Claude on benchmarks and look at the 30% thinking token cut.

Oct 4, 2026claude

Inside Claude Fable 5: The Beast, the Limiter, the Fallout

Claude Fable 5 is Anthropic's most capable model. Developer reactions split over silent AI research restrictions. Pricing, benchmarks, and what this means.

Oct 4, 2026ai

Grok Build Hype vs Reality: A Look at Real User Reactions

Grok Build is xAI's terminal coding agent with Plan Mode and parallel agents. But real users report silent code changes, pricing gripes, and a benchmark gap.

Older posts