Monitoring your OpenAI requests via Dynatrace, you can get cost analysis and forecast estimation via Davis AI, prompt and completion recording, error tracking, performance metrics of your AI services, and more.
The following attributes are available for GenAI Spans.
gen_ai.completion.0.content
gen_ai.completion.0.content_filter_results
gen_ai.completion.0.finish_reason
gen_ai.completion.0.role
gen_ai.openai.api_base
gen_ai.openai.api_version
gen_ai.openai.system_fingerprint
gen_ai.prompt.0.content
gen_ai.prompt.0.role
gen_ai.prompt.prompt_filter_results
gen_ai.request.max_tokens
gen_ai.request.model
gen_ai.request.temperature
gen_ai.request.top_p
gen_ai.response.model
gen_ai.system
gen_ai.usage.completion_tokens
gen_ai.usage.prompt_tokens
llm.request.type
gen_ai.client.generation.choices
none
gen_ai.client.operation.duration
s
gen_ai.client.token.usage
none
llm.openai.embeddings.vector_size
none