v0.44.0 (2025-08-03)
Feat
- sdk: support multiple span processors (#3207)
- semantic-conentions-ai: add LLMVendor enum to semantic conventions (#3170)
Fix
- langchain: spans dictionary memory leak (#3216)
- openai-agents: use framework's context to infer trace (#3215)
- sdk: respect truncation otel environment variable (#3212)
- anthropic: async stream manager (#3220)
- langchain: populate metadata as span attributes in batch operations (#3218)
- anthropic: various fixes around tools parsing (#3204)
- qdrant: fix qdrant-client auto instrumentation condition (#3208)
- instrumentation: remove param
enrich_token_usageand simplify token calculation (#3205) - langchain: ensure llm spans are created for sync cases (#3201)
- openai: support for openai non-consumed streams (#3155)
[main cfe309d] bump: version 0.43.1 → 0.44.0
58 files changed, 77 insertions(+), 57 deletions(-)