Module gemini 

Re-exports

pub use content::*;

pub use count_tokens::*;

pub use generate_content::*;

pub use generation::*;

pub use stream::*;

Modules

audio

Shared speech, translation and transcription settings from Generating content.md.

content

count_tokens

embeddings

Gemini embedContent and batchEmbedContents wire shapes. Source: Embeddings.md.

files

Sources checked 2026-09-13: https://ai.google.dev/api/files and https://generativelanguage.googleapis.com/$discovery/rest?version=v1beta revision 20260910; googleapis/python-genai types.py commit 9052efddccae9f91e699d9b76e0d5a3e032637f5. ProtoJSON accepts both lowerCamelCase and original snake_case field names.

generate_content

Request and non-streaming response bodies for models.generateContent.

generation

live

Gemini Developer Live BidiGenerateContent wire models, checked 2026-09-13. https://ai.google.dev/api/live googleapis/googleapis generative_service.proto commit efc9e8f560a5f1b9b08b62823bfa7955c656cb74; googleapis/python-genai types.py and _live_converters.py commit 9052efddccae9f91e699d9b76e0d5a3e032637f5. The current reference/SDK include fields absent from the public proto; mldev converter mappings determine Developer API wire placement. Vertex-only options are not added. Optional ProtoJSON null is treated as unset; explicit false/0 remains present. Sibling oneof keys are retained without engine validation.

models

stream

Gemini models.streamGenerateContent wire shapes.

video

Gemini Developer API Veo predictLongRunning wire bodies.

voices

Gemini Voices API (/v1beta/voices), audited 2026-10-01. Source: upstream_docs/gemini/docs/Voices.md and the official Interactions OpenAPI. Bodies and query parameters use snake_case. Voice IDs stay in the URL path. These wire types do not provision voices or implement gateway routing.