<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Opencode on Lynx Tech Blog</title><link>https://blog.lynxflow.co/en/tags/opencode/</link><description>Recent content in Opencode on Lynx Tech Blog</description><generator>Hugo -- gohugo.io</generator><language>en</language><lastBuildDate>Wed, 23 Sep 2026 16:55:00 +0800</lastBuildDate><atom:link href="https://blog.lynxflow.co/en/tags/opencode/index.xml" rel="self" type="application/rss+xml"/><item><title>Same Grok 4.7, Five AI CLIs, Only Two Survive: A Terminal Benchmark</title><link>https://blog.lynxflow.co/en/posts/grok-4-7-cli-terminal-benchmark/</link><pubDate>Wed, 23 Sep 2026 16:55:00 +0800</pubDate><guid>https://blog.lynxflow.co/en/posts/grok-4-7-cli-terminal-benchmark/</guid><description>&lt;img src="https://blog.lynxflow.co/images/grok-4-7-cli-terminal-benchmark-cover-v3.webp" alt="Featured image of post Same Grok 4.7, Five AI CLIs, Only Two Survive: A Terminal Benchmark" /&gt; Bottom line: use codex. It answered in 7.4s, aced all reasoning questions, and finished a tool-calling task in 6.9s — best in every category. The official grok CLI is a stable runner-up. Kimi Code and Claude Code fall over the moment you wire them directly to Grok 4.7 — their injected system prompts are too large for a web-account-pool backend to survive. (Update: fixed at the proxy layer the same night this post went live — both now work. See the follow-up section at the end.)</description></item></channel></rss>