@lotte_verheyden: This is big! The team worked hard so Langfuse stays performant, even when you grow to millions and millions of traces
Summary
Langfuse v4 is released with a re-architected data model for improved performance and new features including full-text search, monitors, and an AI assistant.
View Cached Full Text
Cached at: 08/20/26, 10:56 AM
This is big! The team worked hard so Langfuse stays performant, even when you grow to millions and millions of traces
langfuse.com (@langfuse): Langfuse v4 is live on Langfuse cloud and GA for self-hosters.
v4 is a re-architecture of our data model. It is more performant and enables new features (new observations table, full-text search, filter search bar, monitors, code evaluators, Langfuse assistant)
Similar Articles
langfuse/langfuse
Langfuse open-sources its LLM engineering platform to offer self-hosted tracing, analytics, and evaluation tools for production AI applications.
@langfuse: the trace timeline now fits any trace on one screen. twelve spans or twelve hundred. zoom like a map: scroll to pan, piโฆ
Langfuse introduces a responsive trace timeline that fits any trace on one screen with zoom and pan capabilities, enhancing observability for AI agent traces. This update allows efficient navigation of large traces using map-like interactions.
@langfuse: Evals Evals Evals ๐ฆ๐ฒ๐๐๐ถ๐ป๐ด ๐๐ฝ ๐ฒ๐๐ฎ๐น๐๐ฎ๐๐ผ๐ฟ๐ ๐ถ๐ป ๐๐ฎ๐ป๐ด๐ณ๐๐๐ฒ ๐ต๐ฎ๐ ๐ป๐ฒ๐๐ฒ๐ฟ ๐ฏ๐ฒ๐ฒ๐ป ๐ฒ๐ฎ๐๐ถ๐ฒ๐ฟ
Langfuse has introduced easier setup for evaluators with templates, real production data testing, and interactive variable mapping, along with a new UX and a walkthrough video.
@charles_irl: Last fall, we shared our deep dive on FA4 internals. But we didn't stop at grokking the kernel. Since then, we've been โฆ
A blog post details contributions to FlashAttention-4 to improve its performance for large language model inference, especially for decode-heavy workloads, by adjusting parallelism strategies and supporting irregular memory accesses.
@LangChain: Our data agent now handles roughly 40x the request volume our 3 person data team could manage directly. Now, our data tโฆ
LangChain rebuilt its data stack around an AI agent that handles ~40x the request volume of its 3-person data team, enabling self-serve analysis and shifting the team's focus to models, context, and guardrails.