AgentSkillsCN

gemini-model-docs

从官方Gemini API（Google AI）文档中查询当前Gemini模型的ID，以及其GA版与Preview版的发布状态。在SettingsConfiguration、TranscriptionModels或TTS配置中添加或更新Gemini模型时，或当用户询问如何查找模型名称或文档链接时使用。

SKILL.md
--- frontmatter
name: gemini-model-docs
description: Look up current Gemini model IDs and GA vs Preview status from official Gemini API (Google AI) documentation. Use when adding or updating Gemini models in SettingsConfiguration, TranscriptionModels, or TTS config, or when the user asks where to find model names or documentation URLs.

Gemini Model Documentation URLs

Use this skill when you need the current official model IDs for Gemini (transcription, prompt mode, TTS) or when checking GA vs Preview status. The app uses the Gemini API (generativelanguage.googleapis.com), i.e. Google AI for Developers, not Vertex AI.

Where to find model IDs

Gemini API – Models (transcription / prompt mode)

Primary documentation (same API as the app):

Use these pages to get the model ID (e.g. gemini-2.5-flash, gemini-3-flash-preview) and availability (GA vs preview). Prefer stable/GA IDs over dated preview IDs (e.g. gemini-2.5-flash-preview-09-2025) when a GA ID exists.

Programmatic list: The API exposes all available models at:

code
GET https://generativelanguage.googleapis.com/v1beta/models

Use this to verify current IDs and capabilities at runtime if needed.

TTS (Read Aloud)

TTS model IDs for the Gemini API are documented here:

You’ll find IDs such as gemini-2.5-flash-tts and gemini-2.5-pro-tts. The same base URL is referenced in code (e.g. TranscriptionModels.swift).

Optional: Cloud Text-to-Speech – Gemini TTS documents the same/similar TTS models for Cloud/Vertex; use for cross-check only.

How to use in this repo

  1. •Confirm ID and status: Open the Gemini API doc URL (ai.google.dev) for the model or the models overview, and read the model ID and GA/preview status.
  2. •Update code: Set raw values and apiEndpoint in TranscriptionModels.swift, SettingsConfiguration.swift (PromptModel, TTSModel), and any defaults so they use the current GA or preview ID as appropriate.
  3. •Comments: Keep or add comments that point to the Gemini API doc (e.g. “Current Gemini model IDs: https://ai.google.dev/gemini-api/docs/models”).
  4. •Rebuild: After changes, run bash scripts/rebuild-and-restart.sh.

Quick reference – code locations

  • •Transcription / prompt models: WhisperShortcut/TranscriptionModels.swift, WhisperShortcut/Settings/Shared/SettingsConfiguration.swift (PromptModel).
  • •TTS models: WhisperShortcut/Settings/Shared/SettingsConfiguration.swift (TTSModel).
  • •Endpoints: TranscriptionModel.apiEndpoint, TTSModel.apiEndpoint; base is https://generativelanguage.googleapis.com/v1beta/models/{model-id}:generateContent.

Google AI Developers Forum

For Gemini model topics (IDs, deprecations, outages, TTS/API errors), also check the forum. It often has the first reports of issues and official follow-ups:

Useful when: debugging unexplained API/429/500 errors, checking if a model was deprecated or renamed, or before changing model IDs based only on docs.