Library / SDK
Observability & Evaluation
Weave
1Track 2evaluate 3LLM 4applications 5with 6tracing
the six Ws · specification
W1
Who
Maintained by Weights & Biases, the ML experiment-tracking company.
W2
What
A toolkit for tracing, logging, and evaluating LLM and agent applications with dashboards and scored evaluations.
W3
Where
A Python/TypeScript client logging to the W&B cloud, with self-managed options on paid plans.
W4
When
Reach for it when instrumenting an LLM app for debugging and evaluation.
W5
Why
Ties LLM tracing and evaluation into the broader W&B ecosystem.
W6
With
Integrates with OpenAI, Anthropic, LangChain, and LlamaIndex.
W7
Watch
Maintained by Weights & Biases; open-source client (Apache-2.0) backed by a hosted SaaS (open-core). Traces and prompts go to the W&B cloud by default unless self-managed.
alternatives