vai-workflow-model-shootout

Developers choosing between voyage-4-large, voyage-4, and voyage-4-lite need to understand the quality-cost tradeoff for their specific data and queries. Published benchmarks show aggregate scores, but retrieval quality varies by domain.

Install

vai workflow install vai-workflow-model-shootout

How It Works

1. Parallel search — Run the same query through all three models simultaneously
2. Cost estimates — Get cost data for each model in parallel
3. Compare — An LLM generates a side-by-side comparison report with recommendations

Execution Plan

Layer 1 (parallel):  search_large | search_base | search_lite | cost_large | cost_base | cost_lite
Layer 2:             comparison_report

Example Usage

vai workflow run vai-workflow-model-shootout \
  --input query="What are the best practices for database connection pooling?" \
  --input collection="engineering_docs"

What This Teaches

• Six steps in parallel — all search and cost estimate steps have no inter-dependencies
• Directly demonstrates the shared embedding space by querying the same collection with different models
• The generate prompt explicitly mentions the asymmetric embedding capability
• This is the most compelling demonstration of Voyage AI's cost-saving architecture

License

MIT © 2026 Michael Lynn