[Spike] Gemini LLM support - node.js

[kmudduluru]

Have support for gemini LLM

Similar to OpenAI, Customers are using gemini LLM heavily and would like new relic to support it as well.

Basically looking for gemini support similar to: https://docs.newrelic.com/docs/ai-monitoring/intro-to-ai-monitoring/

Feature request here

[workato-integration]

https://new-relic.atlassian.net/browse/NR-359399

[jsumners-nr]

There is a lot of missing detail here. These LLMs support multiple features. Which ones are we supposed to support?

1. Which models are we supposed to target?
   a. https://ai.google.dev/gemini-api/docs/models/gemini
   b. https://ai.google.dev/gemini-api/docs/models/gemini-v2
   c. https://ai.google.dev/gemini-api/docs/models/experimental-models
2. Which "capabilities" are we supposed to support?
   a. https://ai.google.dev/gemini-api/docs/text-generation?lang=node
   b. https://ai.google.dev/gemini-api/docs/vision?lang=node
   c. https://ai.google.dev/gemini-api/docs/audio?lang=node
   d. https://ai.google.dev/gemini-api/docs/long-context
   e. https://ai.google.dev/gemini-api/docs/code-execution?lang=node
   f. https://ai.google.dev/gemini-api/docs/structured-output?lang=node
   g. https://ai.google.dev/gemini-api/docs/imagen
   h. https://ai.google.dev/gemini-api/docs/function-calling
   i. https://ai.google.dev/gemini-api/docs/document-processing?lang=node
   j. https://ai.google.dev/gemini-api/docs/grounding?lang=node
   k. https://ai.google.dev/gemini-api/docs/model-tuning
   l. https://ai.google.dev/gemini-api/docs/embeddings
3. Do we have any LLM experts that have researched this tool and provided details on how it works relative to the ones we already support? I.e. what are the finer details we need to be aware of during instrumentation? For example, our Bedrock support has been being improved by the LLM team because there are operational details that were left out of the original request and are now being filled in by that team.