v3.23.0

• Add info method to logfire query clients by @hramezani in #1204
• Make query client not experimental by @alexmojaki in #1234

v3.22.1

• Fix OpenAI streaming reasoning by @alexmojaki in #1232
• Handle bytes in google genai messages by @alexmojaki in #1231

v3.22.0

• Add instrument_google_genai by @alexmojaki in #1217
• Refactor user tokens, introduce Logfire client by @Viicos in #981
• Use new endpoint for project creation by @hramezani in #1202

v3.21.2

• Fix importlib resources with auto tracing by @alexmojaki in #1212

v3.21.1

• Fix for new OpenAI Agents SDK by @alexmojaki in #1152

v3.21.0

• Add up counter/histogram metrics in spans by @alexmojaki in #1099
• Fix kwargs in proxy metric instruments, preventing warning about duplicate histograms by @alexmojaki in #1149

v3.20.0

• Add baggage utilities by @dmontagu in #1128
• Handle NonRecordingSpans correctly in OpenAI instrumentation by @alexmojaki in #1145

v3.19.0

• logfire.instrument_aiohttp_server() by @adtyavrdhn in #1131
• Fix handling of pydantic_core.Url and pydantic.AnyUrl by @dmontagu in #1130
• Correct context for MCP logs by @alexmojaki in #1138
• Check empty tokens before making connection attempt by @nagarajRPoojari in #1110
• Pass meter provider to Pydantic AI by @alexmojaki in #1136

v3.18.0

• Upgrade to OpenTelemetry SDK 1.34.0 by @alexmojaki in #1120
• Drop Python 3.8 support by @alexmojaki in #1122

v3.17.0

• LangChain instrumentation via LangSmith by @alexmojaki in #1084
• MCP OTel context propagation by @samuelcolvin in #1103
• Add capfire.get_collected_metrics() by @alexmojaki in #1116