Google Gemini

Generate text using Google Gemini models. Support for multimodal inputs (text, images, video, audio), long context windows (up to 2M tokens), and advanced AI capabilities.

Stream text generation in real-time using Gemini models. Returns chunks of text as they are generated for a responsive user experience.

Count the number of tokens in content before sending to avoid hitting limits. Helps estimate costs and ensure prompts fit within context windows.

Generate embeddings for semantic search, clustering, recommendations, and similarity tasks. Convert text into vector representations.

Generate embeddings for multiple texts at once, more efficient than calling embed_content repeatedly.

Get a list of all available Gemini models including their capabilities, context windows, and supported features.

Get detailed information about a specific Gemini model including capabilities, limits, and supported features.

Analyze images with Gemini vision models. Extract information, answer questions, describe content, detect objects, read text (OCR), and more from images.

Create a cached content for reuse across multiple generate_content calls. Caching reduces costs by up to 75% and improves latency. Use for system instructions, common context, or large documents that will be reused. Cache lives for specified TTL (default 1 hour). Returns cache name to use in generate_content's cachedContent parameter.

Create a batch job for async processing with 50% cost savings. Processes multiple requests asynchronously with up to 24 hour turnaround. Ideal for bulk image generation or text processing.

Check the status of a batch job. Returns current state (PENDING, RUNNING, SUCCEEDED, FAILED, CANCELLED, EXPIRED) and result file info when complete.

List all batch jobs for the current API key. Shows job status, creation time, and result file info.

Download results from a completed batch job. Returns parsed JSONL with responses matched to request keys. For image results, returns base64-encoded image data.

Cancel a running batch job. Only works for jobs in PENDING or RUNNING state.

Delete a batch job and its associated files. Only works for completed, failed, or cancelled jobs.

Batch process multiple files (images, PDFs, videos, audio, text) from URLs with 50% cost savings. Auto-fetches files and converts to base64. Supports any publicly accessible URL (R2, S3, CDN, etc.).

Simple chat interface for Gemini. Just send a message and get a response - no complex formatting needed.