Skip to content

Feature Request: Multi-Provider/Multi-Model Support with Agent-Based Selection #866

@mzazakeith

Description

@mzazakeith

Summary

Enable Praison AI to utilize multiple AI providers or the same Provider wit different models, with an agent that intelligently selects the best provider for each task. This should be available in:

  • Auto Agent Mode: The agent automatically routes requests to the most suitable provider.
  • Process: Such as workflow and sequential, where provider selection can be part of the process logic.

Motivation

  • Cost Efficiency: Use lightweight or cheaper providers for simple tasks, reserving more powerful (and expensive) models for complex queries, thus saving on token costs similar to what Cursor do.
  • Performance Optimization: Match provider/model strengths to task requirements (e.g., code execution, summarization, reasoning, etc.).
  • Flexibility: Users can benefit from the strengths of multiple providers/models without manual switching.

Example Use Cases

  • For a simple math calculation, route to a local or lightweight model.
  • For summarizing a web page, use a provider with strong summarization capabilities.
  • For code execution or tool use, select a provider with those internal tools (e.g., Gemini for code execution or Google Search).

Benefits

  • Reduces operational costs by avoiding overuse of expensive models for trivial tasks.
  • Improves response quality by leveraging provider strengths.
  • Streamlines user experience—no need to manually select providers.

Metadata

Metadata

Assignees

No one assigned

    Labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions