4 Comments
User's avatar
Francisco Javier Arceo's avatar

Actually there’s a new vLLM community project working explicitly on agentic inference. Come checkout the project here: https://github.com/vllm-project/agentic-api

Felipe Martínez Cutillas's avatar

So good. The Ollama vs vLLM vs SGLang breakdown is exactly the comparison I keep looking for. Seeing ease-of-setup and throughput laid out side by side makes the trade-off obvious.

jay's avatar

According to Substack's own feature, 51 percent of this article was generated by AI. When watermarking is widely adopted it willl give more confidence to reported numbers like that. Good feature for enabling XAI