Add comprehensive supported models documentation

[jamiechicago312]

Summary

This PR adds comprehensive documentation for all models supported by OpenHands, pulling information directly from the OpenHands Software Agent SDK repository.

Changes Made

New Documentation

• Created openhands/usage/llms/supported-models.mdx - A comprehensive guide covering:
  • Complete list of verified models (OpenHands, OpenAI, Anthropic, Mistral)
  • Information about unverified models available through LiteLLM
  • Model features documentation (reasoning, vision, tool calling, prompt caching)
  • Configuration examples for different providers
  • Model selection guidelines for different use cases
  • Cost considerations and troubleshooting guide

Navigation Updates

• Updated docs.json - Added the new supported models page to the LLM Configuration section

Key Features

Verified Models Coverage

• OpenHands Hosted: 13 premium models including Claude Sonnet 4.5, GPT-5 Codex, etc.
• OpenAI: 12 verified models including GPT-5 family, o-series, and Codex
• Anthropic: 12 verified Claude models from 3.x to 4.5 series
• Mistral: 3 verified Devstral models optimized for coding

Model Features Documentation

• Reasoning Models: Documents which models support reasoning effort and extended thinking
• Vision Support: Information about image input capabilities
• Tool Calling: Native tool calling support details
• Prompt Caching: Cost-saving caching capabilities

Practical Guidance

• Configuration Examples: Code samples for different providers
• Selection Guidelines: Recommendations for production, development, and specialized tasks
• Cost Management: Tips for optimizing model usage costs
• Troubleshooting: Common issues and solutions

Data Source

The documentation is synchronized with the software-agent-sdk repository, specifically:

• openhands-sdk/openhands/sdk/llm/utils/verified_models.py
• openhands-sdk/openhands/sdk/llm/utils/unverified_models.py
• openhands-sdk/openhands/sdk/llm/utils/model_features.py

This ensures the documentation stays accurate and up-to-date with the actual supported models.

Benefits

1. Centralized Information: Single source of truth for all supported models
2. Developer Experience: Clear guidance on model selection and configuration
3. Cost Optimization: Information to help users choose cost-effective models
4. Feature Awareness: Documentation of advanced model capabilities
5. Troubleshooting: Common issues and solutions in one place

Testing

• Verified all model names match the SDK source code
• Confirmed navigation structure works correctly
• Validated MDX syntax and formatting
• Cross-referenced with existing provider documentation

This documentation fills a gap in the current docs by providing a comprehensive overview of ALL supported models, not just the recommended ones, while maintaining synchronization with the actual SDK implementation.

@jamiechicago312 can click here to continue refining the PR