Visit NinjaChat: https://ninjachat.ai/
In this video, I’ll walk you through the big Codex upgrades: Rust-powered terminal speed, GPT-5 model options, ChatGPT subscription support, a new VS Code extension, cloud-run tasks, and how to plug in other providers like DeepSeek via Requesty/OpenRouter.
—
Key Takeaways:
🚀 Codex CLI is now ported to Rust for much faster terminal performance.
🧠 Works with your ChatGPT subscription and supports GPT-5 High/Medium/Low.
🧩 New Codex VS Code extension with Chat, Agent, and Agent Full Access modes.
☁️ Run tasks in the cloud with auto GitHub repo detection and progress tracking.
⚙️ Settings include reasoning effort, manual context, open-file context, and MCP support.
🔄 Switch models with slash models; use Requesty/OpenRouter and DeepSeek seamlessly.
🔑 API keys go via environment variables; launch VS Code from a terminal to load them.
📉 Snappier and lower memory than Roo; clean edited-files list and focused UI.
📈 Dynamic rate limits; $20 ChatGPT plan is great value, $200 plan is mostly unlimited.
🔗 Comparable to Gemini Code Assist; Codex + GPT-5 tool calls feel well-optimized.
—
Timestamps:
00:00 – Introduction
00:07 – GPT-5 + ChatGPT integration and Rust speed
02:09 – Codex VS Code extension and modes
05:28 – NinjaChat
06:18 – Usage
08:33 – Setup of Codex with Deepseek
10:32 – Ending
source
