L 
Langfuse
Open-source observability, evals and prompt management for LLM apps.
NOT REALLY
why this verdict
Keep it — OpenAI has not replaced it
OpenAI's version overlaps, but does not finish the analytics job, so this one is still worth keeping open.
- A named launch, not a vibe named, unlinked
- Model providers ship their own tracing and eval dashboards. OpenAI, February 1, 2025 — no announcement link recorded yet.
- How much of the job it covers not the job editorial call
- Parts of it. The job still needs the tool to get finished.
- Is there a free way to do it? yes
- 1 of 3 listed replacements have a usable free tier: Helicone.
- What the call is worth nothing to cancel
- No paid entry tier tracked, so there is no subscription to cancel.
Reason composed from the fields above missing: announcement linked recorded: free replacement listed recorded: two or more escape hatches
- Threatened by
- OpenAI
- Since
- February 1, 2025
- List price
- free
- Per year
- —
The backstory
Once an app has agents calling tools calling agents, nobody can debug it from logs. Langfuse traces every span, costs it, and lets you score outputs — self-hosted if your data cannot leave. First-party dashboards only cover one vendor's calls, which is rarely the whole picture.
Escape hatches
L 
LangSmith Deeper if you use LangChain.
smith.langchain.com open_in_new H 
Helicone Simpler proxy-based logging.
helicone.ai open_in_new B 
Braintrust Stronger evaluation tooling.
braintrust.dev open_in_new1 of 3 replacements have a usable free tier.