GeminiKit

Build Intelligent Apps

Everything you need to integrate Google's Gemini AI into your Swift applications

🎨

Image Generation

Imagen 3.0 with multiple aspect ratios and batch support

🎬

Video Generation

Veo 2.0 text-to-video and image-to-video creation

🗣️

Text-to-Speech

Natural voices with multi-speaker dialogue support

🧠

Thinking Mode

Step-by-step reasoning with transparent thought process

🌐

Web Grounding

Real-time Google Search for current information

💻

Code Execution

Generate and run code for computational tasks

🛠️

Function Calling

Build AI agents with parallel function execution

🎥

Multimodal

Process images, videos, audio, and documents

⚡

Streaming

Real-time SSE responses with platform optimization

🔤

Embeddings

High-quality vectors for search and similarity

🔄

OpenAI Compatible

Drop-in replacement for existing OpenAI code

🖥️

CLI Tools

Command-line interface for all features

💬

Chat Sessions

Stateful conversations with history management

📊

Structured Output

JSON mode and schema-enforced responses

🔒

Type Safety

Strongly typed Swift API with full autocomplete

📁

File Management

Upload and manage files for processing

🛡️

Safety Controls

Configurable content filtering and harm categories

📦

Zero Dependencies

Pure Swift with no external packages

                    // Initialize GeminiKit

                    import GeminiKit

                    let gemini = GeminiKit(

                      apiKey: "your-api-key"

                    )

                    // Text Generation

                    let response = try await gemini.generateContent(

                      model: .gemini_2_5_flash,

                      prompt: "Explain quantum computing"

                    )

                    print(response.text ?? "")

                    // Streaming Responses

                    for try await chunk in gemini.streamContent(

                      model: .gemini_2_5_pro,

                      prompt: "Write a story"

                    ) {

                      print(chunk.text ?? "")

                    }

                    // Image Generation

                    let images = try await gemini.generateImages(

                      model: .imagen_3_0_generate,

                      prompt: "Japanese garden sunset",

                      aspectRatio: .landscape_16_9

                    )

                    // Multimodal Analysis

                    let result = try await gemini.generateContent(

                      model: .gemini_2_5_flash,

                      messages: [

                        .user("What's in this image?",

                          .imageData(imageData))

                      ]

                    )

                    // Chat Sessions

                    let chat = gemini.startChat(

                      model: .gemini_2_5_flash

                    )

                    let msg = try await chat.sendMessage(

                      "Hello! How are you?"

                    )

                    // Function Calling

                    let result = try await gemini.generateContent(

                      model: .gemini_2_5_flash,

                      prompt: "Get the weather",

                      tools: [weatherTool],

                      toolConfig: .auto

                    )

                    // Thinking Mode

                    let stream = gemini.streamContent(

                      model: .gemini_2_5_flash_thinking,

                      prompt: "Design a system"

                    )

                    // Shows reasoning steps

                    // Text-to-Speech

                    let audio = try await gemini.generateSpeech(

                      model: .gemini_2_5_flash_tts,

                      text: "Hello world!",

                      voice: .aoede

                    )

                    // Embeddings

                    let embedding = try await gemini.embedContent(

                      model: .textEmbedding004,

                      content: "Semantic search text"

                    )

                    // Returns vector array

                    // Web Grounding

                    let result = try await gemini.generateContent(

                      model: .gemini_2_5_flash,

                      prompt: "Latest AI news",

                      tools: [.googleSearch()]

                    )

                    // Code Execution

                    let result = try await gemini.generateContent(

                      model: .gemini_2_5_flash,

                      prompt: "Calculate fibonacci",

                      tools: [.codeExecution()]

                    )

                    // JSON Mode

                    let result = try await gemini.generateContent(

                      model: .gemini_2_5_flash,

                      prompt: "List 3 colors",

                      generationConfig: .init(

                        responseMIMEType: "application/json"

                      )

                    )

                    // Video Generation (Paid)

                    let videoOp = try await gemini.generateVideos(

                      model: .veo_2_0_generate,

                      prompt: "Sunset timelapse",

                      duration: .seconds_8,

                      aspectRatio: .landscape_16_9

                    )

                    // Video Analysis

                    let analysis = try await gemini.generateContent(

                      model: .gemini_2_5_flash,

                      messages: [

                        .user("Analyze this video",

                          .videoFile("video.mp4"))

                      ]

                    )