Vercel.AI.Google
Google Gemini and Vertex provider.
Classes
Section titled “Classes”| Type | Summary |
|---|---|
| FileStreamPart | A streamed file part. The shared stream model has no file delta, so this carries the bytes. |
| GoogleAssistantPart | An assistant content part. |
| GoogleAssistantTurn | An assistant turn. |
| GoogleBatchApi | Gemini Batch API. Downloads the file-based JSONL results of a finished batch. |
| GoogleEmbeddingModel | Gemini embedding model. One value uses :embedContent; several values use :batchEmbedContents. |
| GoogleFileAssistantPart | An assistant file or reasoning file. |
| GoogleFileUrls | URLs Gemini can read without downloading the bytes. |
| GoogleFileUserPart | A user file. Kind is url, data, text, or reference. |
| GoogleFinishReason | Maps Gemini finish reasons onto the shared finish-reason enum. |
| GoogleGeneratedFile | A generated file, including thought images. |
| GoogleImageModel | Gemini image generation. The call runs through GoogleLanguageModel with the IMAGE response modality, so the google provider options apply, and googleSearch enables Google Search grounding. |
| GoogleInteractionsAgent | An agent to call through the Interactions API: { agent: name } or { managedAgent: name } upstream. |
| GoogleInteractionsFinish | Maps an Interactions API status onto a finish reason. |
| GoogleInteractionsInput | Converts a prompt into an Interactions input array. |
| GoogleInteractionsModel | Language model that posts to the Gemini Interactions API. |
| GoogleInteractionsOutput | Reads text and function calls from an Interactions response. |
| GoogleJson | JSON settings shared by Gemini request builders. |
| GoogleJsonAccumulator | Builds a JSON object from Gemini streaming partialArgs chunks. Concatenated text deltas plus Finalize() match the serialized object. Keys such as proto are stored as ordinary properties. |
| GoogleJsonSchema | Rewrites JSON Schema so Gemini responseJsonSchema accepts it. |
| GoogleLanguageModel | Gemini language model. Posts to models/{id}:generateContent. |
| GoogleMessageOptions | Settings for converting a prompt into Gemini contents. |
| GoogleMessages | Converts prompts into Gemini contents. |
| GoogleModelCapabilities | Which Gemini request features a model id accepts. |
| GoogleModelCapability | Classifies Gemini model ids. Unknown future Gemini ids inherit the newest behavior. |
| GoogleModelPath | Builds the resource path Gemini expects for a model id. |
| GoogleOptions | Gemini settings. |
| GoogleParseContext | Ids and metadata keys used while parsing a Gemini response. |
| GooglePartialArgument | One streamed function-call argument fragment. |
| GooglePartialArgumentsFinal | The completed argument JSON and the suffix still needed to close it. |
| GooglePartialArgumentsUpdate | The object and text produced by one GoogleJsonAccumulator update. |
| GooglePreparedRequest | A generateContent request body plus the warnings and extra headers that go with it. |
| GooglePrompt | Gemini contents plus an optional system instruction. |
| GoogleProvider | Google Gemini provider. |
| GoogleProviderTool | A Gemini provider-defined tool such as Google Search. |
| GoogleRealtime | Maps Gemini Live WebSocket messages. No socket is opened, so this compiles on netstandard2.0. |
| GoogleRealtimeClientSecret | An ephemeral Gemini Live auth token and the WebSocket URL it opens. |
| GoogleRealtimeEvent | One normalized Gemini Live event. |
| GoogleRealtimeModel | Realtime model. Event mapping is local; opening the Live socket is not available on this runtime. |
| GoogleRealtimeSession | Session fields accepted by GoogleRealtime. |
| GoogleRealtimeTool | A realtime function tool. |
| GoogleReasoning | Reasoning text that can carry a thought signature. |
| GoogleReasoningAssistantPart | Assistant reasoning. |
| GoogleRequest | Builds a Gemini generateContent body. |
| GoogleResponse | Parses a Gemini generateContent response. |
| GoogleServiceCollectionExtensions | Registers Google and Vertex. |
| GoogleSpeechCall | Speech request settings beyond the shared text and voice fields. |
| GoogleSpeechInput | Inspects a speech transcript before a model call. |
| GoogleSpeechInspection | Transcript text and whether the voice is a custom voice. |
| GoogleSpeechModel | Gemini text-to-speech. Posts to generateContent with an audio response modality. |
| GoogleSpeechPreparation | A prepared Gemini speech request. |
| GoogleSpeechTranslation | Speech translation configuration for the Gemini Live API. The socket itself is not opened here. |
| GoogleSystemTurn | A system turn. |
| GoogleText | Generated text that can carry a thought signature. |
| GoogleTextAssistantPart | Assistant text. |
| GoogleTextUserPart | User text. |
| GoogleThinking | Maps the shared reasoning option onto Gemini thinking configuration. |
| GoogleTokenUsage | Token usage converted from Gemini usageMetadata. |
| GoogleToolCall | A tool call that records whether the provider executed it. |
| GoogleToolCallAssistantPart | An assistant function or server tool call. |
| GoogleToolContentPart | One multimodal tool-result part. |
| GoogleToolPreparation | Tools and toolConfig prepared for a generateContent request. |
| GoogleToolResult | A provider-executed tool result. |
| GoogleToolResultPart | One tool result. |
| GoogleTools | Prepares Gemini function and provider-defined tools. |
| GoogleToolTurn | A tool-result turn. |
| GoogleTranscriptionModel | Gemini transcription. Unary models post to the Interactions API. |
| GoogleTurn | One rich prompt turn used by the message converter. |
| GoogleUsage | Converts Gemini usageMetadata. |
| GoogleUserPart | A user content part. |
| GoogleUserTurn | A user turn. |
| GoogleVertexAnthropicOptions | Settings for Anthropic on Vertex. |
| GoogleVertexAnthropicProvider | Anthropic on Vertex. Sends the Anthropic Messages body to rawPredict with anthropicversion in the body. Vertex does not accept URL sources, the native output format, or strict tools. |
| GoogleVertexAuth | Header construction for Vertex. Live Application Default Credentials are not contacted. |
| GoogleVertexCloudSpeechModel | Cloud Text-to-Speech text:synthesize for Chirp 3 HD voices. |
| GoogleVertexEndpoints | Vertex host and URL construction. Locations are single DNS labels so they cannot rewrite the host. |
| GoogleVertexGeminiTranscriptionModel | Vertex Gemini transcription through generateContent. |
| GoogleVertexMaasProvider | Vertex MaaS OpenAI-compatible endpoint. |
| GoogleVertexPartnerOptions | Settings for the OpenAI-compatible MaaS and xAI endpoints on Vertex. |
| GoogleVertexProvider | Google Vertex AI provider. |
| GoogleVertexSpeechTranscriptionModel | Cloud Speech-to-Text recognize. |
| GoogleVertexXaiProvider | xAI Grok models hosted on Vertex. |
| GoogleWarning | A warning produced while preparing a Gemini request. |
| VertexOptions | Vertex settings. The default route stays compatible with the regional v1 publisher URL. |
Apache-2.0 · Copyright 2023 Vercel, Inc. · unofficial community port, not an official Vercel product