[Feature]: Add Starveri API provider cost/model recognition

## Feature Description and Proposed Solution

Would AgentOps be open to adding cost/model recognition for Starveri API as an OpenAI-compatible provider?

Starveri API is an OpenAI-compatible GPT/Codex API gateway with public model discovery and prepaid-credit billing. For AgentOps users routing traffic through Starveri, recognizing the provider/model pricing would make LLM cost management more accurate instead of showing unknown or unmapped costs.

Public sources:
- Models and pricing: https://api.starveri.net/models
- Docs: https://api.starveri.net/docs
- OpenAPI JSON: https://api.starveri.net/openapi.json
- LLM crawler file: https://api.starveri.net/llms.txt
- Credits/key buyer path: https://api.starveri.net/credits

Current public model IDs from /models:
- gpt-5.1-codex
- gpt-5.3-codex-spark
- gpt-5.3-codex
- gpt-5.4
- gpt-5.4-mini
- gpt-5.5

The /models endpoint publishes USD pricing per 1M tokens for cached input, input, and output. It also publishes tool pricing for web search and code interpreter usage, and notes that context windows are subject to the limits available in Codex.

I am opening this as a feature request because Starveri is a gateway/provider rather than a model author, so I do not want to guess the preferred provider naming or pricing-map representation.

## Related Problem

AgentOps users can point OpenAI-compatible clients at third-party gateways. If a gateway's model IDs/prices are unmapped, cost analytics can be incomplete or misleading.

## Contribution

I can prepare a source-backed PR if maintainers confirm the preferred place to add provider/model pricing metadata.