4 Comments
User's avatar
Serouj's avatar

Why does this sound like an ad for Chat GPT? The reality is that switching costs between LLMs is zero. There are no network effects and you can even take your data with you to a new LLM. The fact is that LLMs are commoditizing and for the vast majority of tasks, you’ll soon be able to run a capable open source LLM on your laptop (e.g. Gemma version 5). In fact, you already can with a MacBook with 64 gb of memory. The future looks pretty bleak for ChatGPT from this perspective!

Past performance does not indicate future performance. All three parts of this series break this paradigm.

Nicola Pellegrino's avatar

Exceptional content, very deep analysis. Thank you!

Udit Batish's avatar

This is such an insightful post, thanks!

Fabrizio Amadini's avatar

Loving these in-depth analyses, Apoorv. Very insightful content. Thx again for sharing! Fab