Command code added GPT-5.6-Sol to the GOAT plan for $70/mo. Billing's still at full price though — $5 in, $30 out.
Openrouter has the same model at half off right now.
Command code added GPT-5.6-Sol to the GOAT plan for $70/mo
More posts
1.7B tokens burned through
My laptop hasn't closed in a week. Multiple worktrees, multiple subagents running at once, 1.7B tokens burned through. I've been building desktop apps, and manual testing on a real machine is a chore. So here's my fully automated setup. I made a subagent called qa. For each small goal I open a worktree and describe it with the /goal slash command, then append: "when the goal is done, have reviewer check the code, have qa drive the mouse to run end-to-end verification, and fix any bugs until every acceptance check passes." The agent works alone in its worktree: research, plan, execute, code review, real-device check. It runs for hours with no one watching. And if the agent runs in a VM, this machine stays free for me to do other things. I also wrote a pi extension that plays a three-note chime the moment a worktree's goal passes acceptance and ships. Install: pi install npm:pi-done-sound
Qwen3.8-27B is kind of absurd.
A 27B model ranks 7th globally on Artificial Analysis for agentic tasks. Runs on a single 3090 or 4090, plenty for personal use. If you're looking for a local general or office agent, this is the one right now.
Claude appeal rejected.
Won't refund the subscription either. I'm half convinced the appeal email is itself phishing, because I'm pretty sure they log the IP. DHH came to Shenzhen recently and showed how he uses Claude Code over there. That's the safest way now, honestly: run Claude Code on an overseas machine, connect through Tailscale, never use it locally. So I need to rent an overseas box with a residential IP, and register and log into Claude there.
Sharing my current mobile only workflow.
Until late September it was the usual setup: parked in front of my laptop, terminal open, bossing pi around, fanning out subagents, then eyeballing diffs myself and filing the PR. Glued to the chair all day, every day. Late September I moved the whole thing to my phone, Hermes runs the show, pi takes the orders. My back is very grateful. Three weeks in, still no regrets. Stack: Hermes on flagship brains (gpt-6-astra / opus-5.5), pi on mid-tier (ds-v4.1-flash / sonnet / terra) with as few plugins as possible, the two talking over a2a. Hermes reads the issues, slices them into small tasks, spins up a worktree for each one, and dispatches a pi into it. When a pi finishes it pings Hermes, who decides: ship the PR or bounce it back for rework. I get back from a full day of being a tourist, or just wake up, and a dozen PRs are sitting there done. Pretty sweet. The catch is how black-box the whole thing is. My emotional arc has been a rerun of early vibe coding, except back then I at least read the diffs. Now even that goes to a flagship model. Boxes all the way down. If you don't have data anxiety, grok-bot and chatgpt-dot basically ship this out of the box, grok-cli and codex built in, native agent-to-agent comms. Downside: your code lives on their servers, everything's moderated, and one ban takes it all with it.
