Crate genai_rs 

genai-rs

A Rust client library for Google’s Generative AI (Gemini) API using the Interactions API.

Quick Start

use genai_rs::Client;

#[tokio::main]
async fn main() -> Result<(), genai_rs::GenaiError> {
    let client = Client::new(
        std::env::var("GEMINI_API_KEY").expect("GEMINI_API_KEY not set")
    );

    let response = client
        .interaction()
        .with_model("gemini-3-flash-preview")
        .with_text("Hello, Gemini!")
        .create()
        .await?;

    println!("{}", response.as_text().unwrap_or("No response"));
    Ok(())
}

Features

• Fluent Builder API: Chain methods for readable request construction
• Streaming: Real-time response streaming with create_stream()
• Function Calling: Automatic function discovery and execution via macros
• Built-in Tools: Google Search, Code Execution, URL Context
• Multimodal: Images, audio, video, and document inputs
• Thinking Mode: Access model reasoning with configurable levels

API Stability & Forward Compatibility

This library is designed for forward compatibility with evolving APIs:

• #[non_exhaustive] enums: Match statements require wildcard arms (_ => ...)
• Unknown variants: Unrecognized API types are captured, not rejected
• Graceful degradation: New API features won’t break existing code

When Google adds new features, your code continues to work. Unknown content types and tools are preserved for inspection via helper methods like has_unknown().

Module Organization

• Client: Main entry point for API interactions
• InteractionBuilder: Fluent builder for configuring requests
• interactions_api: Helper functions for constructing content
• function_calling: Function registration and execution

Re-exports

pub use errors::GenaiError;

pub use content::Annotation;

pub use content::CodeExecutionLanguage;

pub use content::Content;

pub use content::FileSearchResultItem;

pub use content::GoogleSearchResultItem;

pub use content::Resolution;

pub use content::UrlContextResultItem;

pub use request::AgentConfig;

pub use request::DeepResearchConfig;

pub use request::DynamicConfig;

pub use request::GenerationConfig;

pub use request::InteractionInput;

pub use request::InteractionRequest;

pub use request::Role;

pub use request::SpeechConfig;

pub use request::ThinkingLevel;

pub use request::ThinkingSummaries;

pub use request::Turn;

pub use request::TurnContent;

pub use response::AudioInfo;

pub use response::CodeExecutionCallInfo;

pub use response::CodeExecutionResultInfo;

pub use response::ContentSummary;

pub use response::FunctionCallInfo;

pub use response::FunctionResultInfo;

pub use response::GroundingChunk;

pub use response::GroundingMetadata;

pub use response::ImageInfo;

pub use response::InteractionResponse;

pub use response::InteractionStatus;

pub use response::ModalityTokens;

pub use response::OwnedFunctionCallInfo;

pub use response::UrlContextMetadata;

pub use response::UrlContextResultInfo;

pub use response::UrlMetadataEntry;

pub use response::UrlRetrievalStatus;

pub use response::UsageMetadata;

pub use response::WebSource;

pub use tools::FunctionCallingMode;

pub use tools::FunctionDeclaration;

pub use tools::FunctionDeclarationBuilder;

pub use tools::FunctionParameters;

pub use tools::Tool;

pub use wire_streaming::InteractionStreamEvent;

pub use wire_streaming::StreamChunk;

pub use wire_streaming::StreamEvent;

pub use client::Client;

pub use client::ClientBuilder;

pub use request_builder::ConversationBuilder;

pub use request_builder::InteractionBuilder;

pub use function_calling::CallableFunction;

pub use function_calling::FunctionError;

pub use function_calling::ToolService;

pub use streaming::AutoFunctionResult;

pub use streaming::AutoFunctionResultAccumulator;

pub use streaming::AutoFunctionStreamChunk;

pub use streaming::AutoFunctionStreamEvent;

pub use streaming::FunctionExecutionResult;

pub use streaming::PendingFunctionCall;

pub use multimodal::audio_from_file;

pub use multimodal::audio_from_file_with_mime;

pub use multimodal::detect_mime_type;

pub use multimodal::document_from_file;

pub use multimodal::document_from_file_with_mime;

pub use multimodal::image_from_file;

pub use multimodal::image_from_file_with_mime;

pub use multimodal::video_from_file;

pub use multimodal::video_from_file_with_mime;

Modules

client

content

Core content types for the Interactions API.

errors

function_calling

interactions_api

multimodal

Multimodal content utilities for file loading and MIME type detection.

request

Request types for creating interactions.

request_builder

response

Response types for the Interactions API.

streaming

Streaming types for automatic function calling.

tools

wire_streaming

Streaming types for SSE responses.

Structs

FileError

Error information for failed file operations.

FileMetadata

Represents an uploaded file in the Files API.

ListFilesResponse

Response from listing files.

ResumableUpload

Handle to a resumable upload session.

VideoMetadata

Metadata for video files.

Enums

FileState

Processing state of an uploaded file.

Constants

DEFAULT_CHUNK_SIZE

Default chunk size for chunked uploads (8 MB).