recommend_model
Find AI models for specific tasks like image generation or video creation. Describe your task to get ranked recommendations of suitable models from Fal.ai's collection.
Instructions
Get AI-powered model recommendations for a specific task. Describe what you want to do (e.g., 'generate portrait photo', 'anime style illustration', 'product photography') and get the best-suited models ranked by relevance. Featured models by Fal.ai are prioritized.
Input Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| task | Yes | Description of your task (e.g., 'generate professional headshot', 'create anime character', 'transform photo to watercolor') | |
| category | No | Optional category hint to narrow search | |
| limit | No | Maximum number of recommendations |
Implementation Reference
- The handler function for the 'recommend_model' tool. It extracts task, category, and limit from arguments, calls ModelRegistry.recommend_models, formats the recommendations into markdown text content, and handles errors or empty results.
async def handle_recommend_model( arguments: Dict[str, Any], registry: ModelRegistry, ) -> List[TextContent]: """Handle the recommend_model tool.""" task = arguments.get("task") if not task: return [ TextContent( type="text", text="❌ Please describe your task (e.g., 'generate professional headshot').", ) ] category = arguments.get("category") limit = arguments.get("limit", 5) recommend_result = await registry.recommend_models( task=task, category=category, limit=limit ) recommendations = recommend_result.recommendations if not recommendations: return [ TextContent( type="text", text=f"No models found for task: '{task}'. Try a different description or remove the category filter.", ) ] # Format output lines = [f'## Recommended Models for: "{task}"\n'] if recommend_result.used_fallback: lines.append( f"⚠️ *Using cached results ({recommend_result.fallback_reason}). Results may be less relevant.*\n" ) lines.append("💡 *Models are ranked by relevance. ⭐ = Featured by Fal.ai*\n") for i, rec in enumerate(recommendations, 1): # rec is a dictionary from model_registry.recommend_models model_id = rec.get("model_id", "unknown") name = rec.get("name") description = rec.get("description") highlighted = rec.get("highlighted", False) group = rec.get("group") score = rec.get("score", 0.0) # Badge for highlighted models highlighted_badge = " ⭐" if highlighted else "" lines.append(f"### {i}. `{model_id}`{highlighted_badge}") if name: lines.append(f"**{name}**") if description: lines.append(f"{description}") if group: lines.append(f"*Family: {group}*") lines.append(f"*Relevance: {score:.1%}*\n") return [TextContent(type="text", text="\n".join(lines))] - The Tool schema definition for 'recommend_model', including inputSchema with required 'task' parameter and optional 'category' and 'limit'.
name="recommend_model", description="Get AI-powered model recommendations for a specific task. Describe what you want to do (e.g., 'generate portrait photo', 'anime style illustration', 'product photography') and get the best-suited models ranked by relevance. Featured models by Fal.ai are prioritized.", inputSchema={ "type": "object", "properties": { "task": { "type": "string", "description": "Description of your task (e.g., 'generate professional headshot', 'create anime character', 'transform photo to watercolor')", }, "category": { "type": "string", "enum": ["image", "video", "audio"], "description": "Optional category hint to narrow search", }, "limit": { "type": "integer", "default": 5, "minimum": 1, "maximum": 10, "description": "Maximum number of recommendations", }, }, "required": ["task"], }, ), - src/fal_mcp_server/server.py:61-85 (registration)Registration of the 'recommend_model' handler in the TOOL_HANDLERS dictionary used by the MCP server to route tool calls. The handler is imported earlier in the file.
TOOL_HANDLERS = { # Utility tools (no queue needed) "list_models": handle_list_models, "recommend_model": handle_recommend_model, "get_pricing": handle_get_pricing, "get_usage": handle_get_usage, "upload_file": handle_upload_file, # Image generation tools "generate_image": handle_generate_image, "generate_image_structured": handle_generate_image_structured, "generate_image_from_image": handle_generate_image_from_image, # Image editing tools "remove_background": handle_remove_background, "upscale_image": handle_upscale_image, "edit_image": handle_edit_image, "inpaint_image": handle_inpaint_image, "resize_image": handle_resize_image, "compose_images": handle_compose_images, # Video tools "generate_video": handle_generate_video, "generate_video_from_image": handle_generate_video_from_image, "generate_video_from_video": handle_generate_video_from_video, # Audio tools "generate_music": handle_generate_music, } - Core helper method in ModelRegistry that performs semantic search based on the task, builds recommendations with metadata and reasoning, and handles fallback to cached results.
async def recommend_models( self, task: str, category: Optional[str] = None, limit: int = 5, ) -> RecommendationResult: """ Recommend the best models for a given task. Uses semantic search and prioritizes highlighted (featured) models. Returns models with relevance reasoning. Args: task: Description of the task (e.g., "generate professional headshot") category: Optional category hint ("image", "video", "audio") limit: Maximum number of recommendations Returns: RecommendationResult with recommendations and fallback indicator """ # Map simplified category to API category if provided api_category = None if category: # Use primary mapping for each simplified category category_to_api = { "image": "text-to-image", "video": "text-to-video", "audio": "text-to-audio", } api_category = category_to_api.get(category) # Search using the task as query search_result = await self.search_models( query=task, category=api_category, limit=limit * 2 ) # Build recommendations with reasoning recommendations: List[Dict[str, Any]] = [] for model in search_result.models[:limit]: rec = { "model_id": model.id, "name": model.name, "description": model.description, "category": model.category, "highlighted": model.highlighted, "group": model.group_label, "reason": self._generate_recommendation_reason(model, task), } recommendations.append(rec) return RecommendationResult( recommendations=recommendations, used_fallback=search_result.used_fallback, fallback_reason=search_result.fallback_reason, )