In this episode, @HenrikRexed is asking a critical question: How much energy does your AI prompt really consume? ⚡🌍 As LLMs like ChatGPT, Claude, and Gemini become part of our daily workflows, it's time to look beyond performance and ask: what’s the environmental cost of all this compute? In this episode, we explore:
- 🔍 The fundamental differences between traditional apps and LLMs
- 📊 How to observe LLM workloads using tools like OpenLLMetry and OpenLit
- 🌱 How to estimate energy usage and emissions with the open-source project Ecologits
- 🖥️ When to use CodeCarbon for self-hosted LLMs
- 📈 How to export energy data as logs or OpenTelemetry metrics
---------------------------
📖 Chapters 📖
00:00 Intro
02:20 Differences between traditional apps and LLMs
07:45 LLM Observability
13:11 Ecologits
19:48 Conclusion
-----------------------------
🔗 Additional Links
Github repo for the tutorial
ecolotigits
ecologits algorithm
Pr on the Otel support
OpenLLmetry
OpenLit
CodeCarbon
Blog
Dynatrace Trial
-----------------------------
Subscribe to our YT channel
Stay up-to-date with Dynatrace! Follow us on Facebook, Instagram, LinkedIn, Twitter, Twitch
