[Feature]: Enhance Model Info Lookup for litellm_proxy Prefixed Models

[openhands-agent]

What problem or use case are you trying to solve?
The current model info lookup using model_name gets confused with litellm_proxy prefixed models like litellm_proxy/bedrock/us.anthropic.claude-3-5-sonnet-20241022-v2:0, which leads to incorrect model defaults and no prompt caching. There's a need to map this to the correct model name to load the proper 'Model info'.

Describe the UX of the solution the user wants
The user wants a straightforward way to look up model information without manually stripping prefixes or configuring model names.

User's thoughts on the technical implementation?
User noticed the lookup occurs in llm.py. A potential solution involves modifying to accommodate the litellm_proxy prefix, possibly by removing it before processing, as seen in user-specific code changes.

Describe the alternatives considered
Current workaround involves modifying llm.py to strip the prefix manually, but this isn't ideal for all scenarios.

Additional context

• Slack thread for additional discussion and logs: https://openhands-ai.slack.com/archives/C06U8UTKSAD/p1737397675586899?thread_ts=1737397675.586899&cid=C06U8UTKSAD
• Related code snippets: OpenHands GitHub Link

Issue Created By: Steve Cosenza on Slack

Link to Slack Thread: Slack Link