OpenRouter Model Selector icon

OpenRouter Model Selector

Intelligently select the optimal OpenRouter model based on task, budget, and benchmarks

OpenRouter Model Selector

AI-generated

Summary

Selects and ranks OpenRouter AI models based on a specified task category, budget, and optional filters, returning the best matching model and alternatives with detailed scoring and reasoning.

Inputs

  • Task Category (required) — Specifies the type of task the model will perform, such as translation, coding, chat, vision, embedding, summarization, analysis, or math.
  • Budget (required) — Sets the spending preference affecting model selection, choosing between 'cheap', 'balanced', and 'premium' options.
  • Model Override — Optionally overrides automatic model selection by specifying a particular model ID from the top models for the chosen task and budget.
  • Filters — Optional filter collection to restrict models by minimum context length, JSON mode support, vision capability, maximum cost per 1K tokens, provider whitelist, and blacklist.
  • Options — Additional options including the number of alternative models to return and the output detail level ('full', 'simple', or 'compact').

Output shape

a single JSON object containing the recommended model and alternatives based on the chosen output format (full, simple, or compact)

Returns model details including scores, pricing, capabilities, and reasoning when 'full' or 'simple' output formats are selected. In 'compact' mode, returns only model IDs. Automatic model selection calculates a composite score from benchmarks, task fit, budget fit, and capabilities. A manual override returns the specified model without scoring. Errors occur if no suitable models are found or if a blocked model is selected.

Examples

Example 1: User wants to select the best OpenRouter model for conversational chat with a balanced budget, default filters, and desires three alternatives in full detail.

Task Category: chat, Budget: balanced, Model Override: default, Filters: default, Options: Alternatives Count = 3, Output Format = full

Example 2: User needs a low-cost model for data analysis requiring JSON output mode, limits providers to OpenAI and Anthropic, and wants only model IDs as output.

Task Category: analysis, Budget: cheap, Filters: Require JSON Mode = true, Provider Whitelist = [OpenAI, Anthropic], Options: Output Format = compact

Links

Discussion