Pi 2.0 (Codemode, Durable Tools & More!): Native via Codemode, Custom Model Routing & Pi Durable
In this video, I'll go through the Pi 1.0 release from Earendil and the new Pi Durable package. We'll start with native MCP support through Codemode, where the agent writes small JavaScript programs that call tools, combine results, and send back only what's useful, plus the specialist models it can reach, like the Jev classifier and image generation. Then I'll cover virtual models for custom model routing, deferred tool loading, cache warming, mid-conversation system messages, the new system theme, and fullscreen mode becoming the default, along with the smaller changes in the final release. Finally, I'll explain Pi Durable for long-running agent applications, how to install Pi, and what it actually costs to run.--
Key Takeaways:
🚀 Pi 1.0 shipped on October 1 with native MCP support through Codemode, custom model routing, deferred tool loading, cache warming, and a refreshed terminal interface.
🧠 Codemode lets the agent write a small JavaScript script that calls tools in parallel and returns only the useful output, so intermediate data stays out of the model's context.
⚠️ Scripts run in a sandbox, but the tools they call keep their own permissions, and tool calls made before a failure are not undone.
🔍 Jev answers typed questions with probabilities and image models generate images through Codemode, but generated images are not saved to disk unless your workflow saves them.
🔀 Virtual models let an extension route each request to a different physical model, and Pi shows the routed model and cost per model in /session.
💸 Routing has tradeoffs: switching models can lose prompt-cache reuse and a classifier adds latency, so compare against a single-model baseline on real tasks.
⚡ Cache warming only refreshes when Pi estimates at least five cents in avoided cache-miss cost, and the refresh usage counts toward your session totals.
💻 The new system theme builds Pi's colors from your terminal's palette, and fullscreen is now the default, with regular mode available through /settings or --tui-mode regular.
📊 The team reports a GPT-5.6 request with default tools and Codemode drops from about 5,300 to 3,300 prompt tokens, which is about 40% in that specific setup.
✅ Pi Durable is an experimental, MIT-licensed foundation for long-running agent apps with checkpoints and resumable tasks; it's a strong release if you like shaping your own coding agent. Receive SMS online on sms24.me
TubeReader video aggregator is a website that collects and organizes online videos from the YouTube source. Video aggregation is done for different purposes, and TubeReader take different approaches to achieve their purpose.
Our try to collect videos of high quality or interest for visitors to view; the collection may be made by editors or may be based on community votes.
Another method is to base the collection on those videos most viewed, either at the aggregator site or at various popular video hosting sites.
TubeReader site exists to allow users to collect their own sets of videos, for personal use as well as for browsing and viewing by others; TubeReader can develop online communities around video sharing.
Our site allow users to create a personalized video playlist, for personal use as well as for browsing and viewing by others.
@YouTubeReaderBot allows you to subscribe to Youtube channels.
By using @YouTubeReaderBot Bot you agree with YouTube Terms of Service.
Use the @YouTubeReaderBot telegram bot to be the first to be notified when new videos are released on your favorite channels.
Look for new videos or channels and share them with your friends.
You can start using our bot from this video, subscribe now to Pi 2.0 (Codemode, Durable Tools & More!): Native via Codemode, Custom Model Routing & Pi Durable
What is YouTube?
YouTube is a free video sharing website that makes it easy to watch online videos. You can even create and upload your own videos to share with others. Originally created in 2005, YouTube is now one of the most popular sites on the Web, with visitors watching around 6 billion hours of video every month.
