{"type":"rich","version":"1.0","provider_name":"Transistor","provider_url":"https://transistor.fm","author_name":"Lenny’s AI Builders","title":"i let OpenAI's new Terra drive my computer. then Opus 5.5 got the same job | LAB 0030","html":"<iframe width=\"100%\" height=\"180\" frameborder=\"no\" scrolling=\"no\" seamless src=\"https://share.transistor.fm/e/1a89494b\"></iframe>","width":"100%","height":180,"duration":835,"description":"OpenAI just dropped GPT-6.1 Terra - a model that drives your computer, at half the price of GPT-6.1 Sol. so i let it do the admin in a made-up plumbing office on my computer: find the enquiry, book it in, write the quote. then i gave Opus 5.5 the exact same job, against the clock. both got it right, the faster one isn't the one i'd hire, and the price surprised me.\n\nalso this week: Theo says the Claude Agent SDK drains your 5-hour limit faster than Claude Code (a GitHub issue a month ago measured about 1.8x per token, and Theo reports weekly limits aren't affected), OpenAI says its models proved three big results (nobody outside OpenAI has signed off yet), Anthropic filed to go public (revenue in the first half of 2026 was 14x the first half of 2025: $16.3 billion vs $1.2 billion), and AI video editing: CapCut's plugin for ChatGPT (announced), Instagram testing long horizontal videos, and Riverside's agents (live).\n\nthe practical - BuilderBench, my own benchmark: GPT-6.1 Terra vs Kimi K3.1 Pro on 4 hours of real Threadify jobs, scored blind. one category wasn't even close. and a prompt that runs a mini blind test of any two models on one of your own jobs. save each model's answer in its own folder inside a folder called outputs, then paste this into Claude Code in the folder that holds outputs.\n\nthe prompt:\nMake me a blind model test for my own work, like a mini BuilderBench.\n\n1. Look in the folder ./outputs. Each subfolder is one model's work on the same job (the folder name is the model's name).\n2. Copy each model's work into ./blind/ under a random label (Option A, Option B, ...). Save which label is which model in ./sealed/key.json. Don't show me the key and don't put model names anywhere in ./blind/.\n3. Build one page, ./blind/score.html, that shows every option side by side (open HTML files in an iframe, show text files as text). Under each option: a score out of 10, one line \"would I use this as is? yes / with fixes / no\", and a notes box.\n4. Add a \"Lock...","thumbnail_url":"https://img.transistorcdn.com/P0B9fKUWl01KOkVXqQcDv5YBEKR1sflq-Hk71z-fa_U/rs:fill:0:0:1/w:400/h:400/q:60/mb:500000/aHR0cHM6Ly9pbWct/dXBsb2FkLXByb2R1/Y3Rpb24udHJhbnNp/c3Rvci5mbS84MzAx/ZGNmYzY1MmEwNjBk/NzUzN2NmODM3ZWUx/YzMzZi5wbmc.webp","thumbnail_width":300,"thumbnail_height":300}