Select Model Only
AI-generatedSummary
Select and rank language models suitable for a specified task and budget without making API calls, returning the top recommended model plus alternatives.
Inputs
- Task Category (required) — The type of task the model will perform, such as chat, coding, translation, vision, etc.
- Budget (required) — Spending preference level: cheap, balanced, or premium.
- Domain — Optional domain string to apply specific routing policy overrides.
- Model Override — Manually specify a particular model to return instead of auto-selection.
- Filters — Filters to narrow models by minimum context length, JSON mode support, vision capability, maximum cost per 1K tokens, and provider whitelisting/blacklisting.
- Options: Alternatives Count — Number of alternative ranked models to return alongside the recommended selection (defaults to 3).
Output shape
an object containing 'recommended' model details plus an array of 'alternatives', and optional routing metadata.
The output includes comprehensive scoring details for each model, including score breakdown and human-readable reasoning. If a model override is used, alternatives are empty. Models are filtered and scored based on benchmarks, task fit, budget compliance, and capabilities using data fetched from a Supabase model catalog. The operation does not perform any API chat calls, only model selection and ranking.
Examples
Example 1: Select the best models for conversational chat with a balanced budget and default filters, returning 3 alternatives.
Task Category: Conversational Chat; Budget: Balanced; Alternatives Count: 3; no domain or manual model override.