Generate embeddings for multiple texts in a batch operation. This tool efficiently generates embedding vectors for multiple texts at once, optimized for bulk embedding tasks with a single API call. Args: - model (string): Embedding model (default: gemini-embedding-001) - texts (array): Array of texts to embed (max 100 items) - taskType (string, optional): Optimization hint - response_format ('json' | 'markdown'): Output format Returns: Array of embedding vectors corresponding to input texts.
Multi-turn conversation with Gemini models. This tool supports multi-turn conversations by accepting a message history. Each message has a role ('user' or 'model') and content. Args: - model (string): Gemini model ID (default: gemini-2.5-flash) - messages (array): Conversation history - role ('user' | 'model'): Message sender - content (string): Message text - systemInstruction (string, optional): System instruction - generationConfig (object, optional): Generation parameters - response_format ('markdown' | 'json'): Output format Returns: Model's response to the conversation. Examples: - Single turn: [{ role: "user", content: "Hello!" }] - Multi-turn: Include previous user and model messages
Count tokens in text for a specific model. This tool estimates how many tokens a given text would consume when processed by a specific Gemini model, useful for managing token budgets. Args: - model (string): Model for tokenization (default: gemini-2.5-pro) - text (string): Text to count tokens for - response_format ('json' | 'markdown'): Output format Returns: Token count information.
Generate embedding vectors for text using Gemini embedding models. This tool converts text into numerical vectors for semantic search, clustering, classification, and similarity comparison. Args: - model (string): Embedding model (default: gemini-embedding-001) - text (string): Text to embed - taskType (string, optional): Optimization hint - RETRIEVAL_QUERY: For search queries - RETRIEVAL_DOCUMENT: For documents to be searched - SEMANTIC_SIMILARITY: For similarity comparison - CLASSIFICATION: For classification tasks - CLUSTERING: For clustering tasks - outputDimensionality (number, optional): Vector dimensions - response_format ('json' | 'markdown'): Output format Returns: Embedding vector as array of floating-point numbers. Examples: - Embed "machine learning" for search query - Use RETRIEVAL_DOCUMENT for indexing documents
Generate text content using Google Gemini models. This tool sends a text prompt to Gemini and returns the model's response. Supports various generation parameters like temperature, top-p, and max tokens. Args: - model (string): Gemini model ID (default: gemini-2.5-flash) - prompt (string): Text prompt to send to the model - systemInstruction (string, optional): System instruction to guide behavior - generationConfig (object, optional): Generation parameters - temperature (number): Randomness control (0.0-2.0) - topP (number): Nucleus sampling (0.0-1.0) - topK (number): Top-K sampling - maxOutputTokens (number): Maximum response length - candidateCount (number): Number of responses (1-8) - stopSequences (string[]): Stop generation sequences - response_format ('markdown' | 'json'): Output format Returns: Generated text content from the model. Examples: - "Explain quantum computing" -> Generate explanation - "Write a poem about spring" -> Creative writing - Use temperature=0.0 for factual responses - Use temperature=1.0+ for creative content
Get detailed information about a specific Gemini model. This tool retrieves comprehensive metadata for a single model, including its capabilities, token limits, and configuration options. Args: - model (string): Model name (e.g., 'gemini-2.5-flash') - response_format ('markdown' | 'json'): Output format Returns: Detailed model information.
List available Gemini models with pagination support. This tool retrieves information about available Gemini models, including their capabilities, token limits, and supported methods. Args: - pageSize (number): Number of models per page (1-1000, default: 50) - pageToken (string, optional): Token for pagination - response_format ('markdown' | 'json'): Output format Returns: List of available models with their metadata.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 8 | 2025-06-18+ | v2 |
| 2026-03-09 | F | 0 | - | v1 |