| name | or-find-tool-models |
| description | Use when the user wants to find OpenRouter models that support tool use / function calling. Triggers on phrases like "OpenRouter models with tool use", "which OR models support function calling", "find tool-calling models on OpenRouter", "MCP-compatible models on OpenRouter", "models for agentic workflows on OR". |
Find OpenRouter Tool-Calling Models
Filter the OpenRouter catalog to models that declare tool/function calling support, then rank or summarize them based on the user's stated criteria (price, context, provider).
When to use
The user wants to discover models suitable for agentic / tool-using workflows on OpenRouter — function calling, MCP, code agents, structured tool invocation.
Workflow
- Fetch the catalog:
curl -s https://openrouter.ai/api/v1/models -H "Accept: application/json"
- Filter
data[] where supported_parameters includes "tools". Optionally also require "tool_choice" for stricter filtering.
- If the user expressed preferences (cheapest, largest context, specific provider), apply them:
- Cheapest — sort by
pricing.completion ascending
- Largest context — sort by
context_length descending
- Provider — filter
id by prefix (openai/, anthropic/, google/, meta-llama/, etc.)
- If the user did not state criteria, ask one short clarifying question: budget ceiling? minimum context? preferred providers?
- Return the top 5–10 matches as a markdown table: ID, Context, Prompt $/1M, Completion $/1M, Other supported params.
Notes
- Prices in the catalog are USD per token. Multiply by 1,000,000 for $/1M-token display.
- Some models list
tools but have caveats noted in their description — surface those if relevant.
- Free models (pricing all
"0") often have rate limits and lower reliability — flag this when recommending free models.
Output
Markdown table sorted by the user's stated criterion (default: cheapest completion price). Include a one-line summary of how many tool-capable models exist in the catalog total.
