Stories by bisonbear

I ran Sol 6 vs. 6.1 on tasks from my own repo. It's cheaper and at least as good

I compared Opus 4.8 vs. Opus 5 on 25 of my tasks to see what the difference was

I compared 5 popular token saving methods in Codex and found that none delivered

I ran Sonnet 5 vs. Opus 4.8 head to head on 24 tasks to see what's different

I evaluated GLM 5.2 against the frontier on tasks from real repos

I benchmarked Opus 4.8 vs. GPT 5.5 on 2 open source repos

I used autoresearch to improve my AGENTS.md, measured against real tasks

A brief investigation into the GPT-5.5 regression claims

The Opus 4.7 reasoning curve - Medium is the best default?

GPT-5.5 low vs. medium vs. high vs. xhigh: the reasoning curve on 26 real tasks

GPT-5.5 vs. GPT-5.4 vs. Opus 4.7 on 56 real coding tasks from 2 open source repo

I ran Opus 4.7 vs. Old Opus 4.6 vs. New Opus 4.6 on 28 Zod tasks

Coding evals are broken. CI is green while AI code quality goes unmeasured

Agents.md is the highest-leverage code you're not testing

Your AI coding benchmark is hiding a 2x quality gap

Things I Learned at the Claude Code NYC Meetup

Claude vs. Codex in the Messy Middle

Spacetime as a Neural Network

One agent isn't enough

Context Engineering: The New Skill for Working with AI Agents