Claude Code is like one very competent elf warrior, great with a blade, excellent inside one battle/repo.
Codex is like a dwarf with a hammer and shell access, less poetry, and more of “I fixed the wall and ran the tests.”
Hermes is Gandalf, not swinging himself, but sending people on quests.
Spent the evening testing Qoder and Qoder Work by @Alibaba_Qwen and it's actually insane. Not just another agentic coder, it's got its agentic UI and a VS fork like Cursor, plus a built-in Repo Wiki or a knowledge tab, that acts like Obsidian's second brain for your codebase... They merged all our tools into one app..
And it's got persistent memory that learns as you work too, fast computer use, browser use and a model picker with Qwen, GLM, DeepSeek, Kimi, MiniMax...
They really cooked with both of these desktop apps..
@bcherny one ask as a daily Claude Code user: unified persistent memory across surfaces. Right now CC, chat, and the browser extension don't share context - I'm the connective tissue. Is a single synced memory layer on the roadmap?
Gemini’s privacy tradeoff is insane.
If I keep chat history on, Google says some chats may be reviewed by humans and used to improve/train AI.
If I turn it off, chats only stick around for 72 hours.
Why is “save my chat history” bundled with “let humans review my chats”?
@OfficialLoganK@GeminiApp
Claude Code is like one very competent elf warrior, great with a blade, excellent inside one battle/repo.
Codex is like a dwarf with a hammer and shell access, less poetry, and more of “I fixed the wall and ran the tests.”
Hermes is Gandalf, not swinging himself, but sending people on quests.
Quick Q on the new @NousResearch Hermes desktop app.
Why does it show TWO Anthropic options? "Anthropic Claude" (OAuth) vs "Claude Code" (CLI token).
Which bucket does each bill, Agent SDK credit vs Claude Code subscription quota? And is the Claude Code one ToS-safe given the April ban?