Major Releases: TypeScript for LLM Evals, Total Fidelity ML Metrics, & More
Spring is in the air, and we’re excited to bring you four fresh releases in the Comet platform to make your AI tracking and analysis flows more accessible and precise.
For JavaScript developers building with LLMs, Opik’s new TypeScript SDK brings our full suite of LLM tracing, evaluation, and observability tools into a familiar language for faster iteration. And if you’re using OpenTelemetry, our new integration makes it easy to add Opik to your application and connect frameworks that support OpenTelemetry observability.
In Comet Experiment Management, teams building and training ML models at massive scale can now store metrics with total fidelity — that means zero sampling, with the ability to drill down and examine details on experiments with millions of steps. Plus, working with sound just got easier — new audio panels in your dashboard let you listen to and analyze generated audio files side-by-side with their spectrograms, all in the same tab.
Last, a huge thank you to our community of Comet users for making Opik one of the fastest growing LLM eval frameworks on GitHub, with 5,000 stars and counting. If you haven’t started building your AI applications with Opik, check out the repo or try the free hosted version.
Read on to see how these updates streamline your LLM and ML development workflows!
Opik: TypeScript SDK
The Opik TypeScript SDK is now live! AI is no longer solely for data scientists; engineers are now building AI-powered products each day, and now the Opik TypeScript gives you a seamless way to integrate AI into your workflow. You can now evaluate, test, and monitor your AI applications directly from your JavaScript/TypeScript codebases with ease.
Opik x OpenTelemetry Integration
It’s now easier than ever to add Opik to applications that use OpenTelemetry, and connect frameworks that support OpenTelemetry observability. This new integration lets you use any OpenTelemetry SDK to send traces directly to Opik, and extends Opik’s language support beyond the official Python and TypeScript SDKS.
ML Experiment Management: Total Fidelity Metrics
Need to drill down to the details of an absolutely massive experiment with millions of steps? Now you can store metrics without sampling for uncompromising granularity at scale. Add a Python panel to visualize the total fidelity metric of your choice, or use an SDK method to build your analysis of the full data.
ML Experiment Management: Audio Panels
Audio Panel capabilities are now live! In addition to our Image & Video panels, you can now analyze generated audio files side-by-side with audio spectrograms, allowing you to listen to samples while examining the spectrograms, removing the manual task of switching back and forth within experiment tabs.
How to Upgrade
If you are using www.comet.com, no action is required! You have access to these features and can use them immediately.
If you are using a self-hosted version of Comet, before you upgrade, please verify which version of Comet you are running.
Comet Versions March 2025
Frontend version: 5.92.26
Backend version: 3.59.279
Optimizer version: 2.1.4
SDK version: 3.49.1
To check your frontend version: Click your username in the top-right corner of the Comet application and check the Version at the bottom of the pop-up.
To check your backend version: Visit. <your_comet_url>.com/api/isAlive/ver
To check your SDK version: In your training environment, run the command pip freeze | grep comet_ml
If you would like to upgrade, please contact your Technical Account Manager or Deployment Engineer on the Comet team. You can always reach out directly to Chase Fortier (chasef@comet.com) with general questions.
Related Articles
