Package com.tnsai.llm.providers

Class VertexAIClient

java.lang.Object

com.tnsai.llm.providers.AbstractLLMClient

com.tnsai.llm.providers.VertexAIClient

All Implemented Interfaces:

com.tnsai.llm.LLMClient

public class VertexAIClient
extends AbstractLLMClient

Google Cloud Vertex AI LLM client (enterprise Gemini).

Vertex AI is the enterprise variant of the Gemini API — same underlying models, but hosted in customer-owned GCP projects with GCP IAM auth, regional quotas, and enterprise SLAs. The wire format is the native Gemini generateContent shape (contents[].parts[], systemInstruction, generationConfig) — distinct from the OpenAI-shaped providers in the rest of this module.

The endpoint URL is regional and project-scoped: https://{LOCATION}-aiplatform.googleapis.com/v1/projects/{PROJECT}/locations/{LOCATION}/publishers/google/models/{MODEL}:generateContent. Both LOCATION and PROJECT are caller-supplied; this client derives the base URL from the configured location when no explicit VERTEX_AI_BASE_URL is provided.

Supported models

• gemini-2.0-flash-001 — frontier multimodal flash tier
• gemini-2.0-pro-001 — higher-quality general purpose
• gemini-1.5-pro-002 — long-context (1M-2M token) tier
• gemini-1.5-flash-002 — cheap / fast

The catalog evolves; check cloud.google.com/vertex-ai/generative-ai/docs/model-reference/gemini.

Auth (v1 scope)

v1 takes a pre-fetched OAuth access token via VERTEX_AI_API_KEY. Two practical sources:

• Development: gcloud auth print-access-token — gives a 1-hour-lived token tied to the user's gcloud login.
• Production: a sidecar / workload-identity helper that refreshes the token from a service-account JSON file (or GKE metadata server) and writes it where this client can read it.

The framework does not yet implement Application Default Credentials (ADC) end-to-end — that's a follow-up issue (service-account JWT signing, GCE metadata-server probe, gcloud config fallback) because it pulls in google-auth-library-java or hand-rolled JWT crypto. Until then, callers manage token refresh themselves.

Configuration via environment variables:

• VERTEX_AI_API_KEY — Required. A pre-fetched OAuth access token (see above).
• VERTEX_AI_PROJECT_ID — Required. The GCP project ID hosting the Vertex AI endpoint.
• VERTEX_AI_LOCATION — Optional. The GCP region (us-central1, europe-west4, etc.). Defaults to us-central1.
• VERTEX_AI_BASE_URL — Optional. Fully overrides the host (useful for proxies / private endpoints). If unset, derived from VERTEX_AI_LOCATION.

See Also:

• Vertex AI REST reference

Constructor Summary

Constructors

Constructor	Description
VertexAIClient(String model)
VertexAIClient(String model, float temperature)
VertexAIClient(String model, float temperature, Float topP, Integer maxTokens)
VertexAIClient(String model, float temperature, Integer maxTokens)
VertexAIClient(String baseUrl, String model, float temperature, Float topP, Integer maxTokens, String apiKey, String projectId, String location)

Method Summary

Modifier and Type	Method	Description
com.tnsai.llm.ChatResponse	chat(String message, Optional<String> systemPrompt, Optional<List<Map<String,Object>>> history, Optional<List<Map<String,Object>>> tools)
Stream<String>	streamChat(String message, Optional<String> systemPrompt, Optional<List<Map<String,Object>>> history, Optional<List<Map<String,Object>>> tools)

Methods inherited from class com.tnsai.llm.providers.AbstractLLMClient

getBaseUrl, getCallTimeoutMs, getMaxTokens, getModel, getTemperature, getTopP

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Methods inherited from interface com.tnsai.llm.LLMClient

chat, chat, chat, chat, chat, getCapabilities, getProviderName, listModels, streamChat, streamChat, streamChatWithHandler, streamChatWithHandler, streamChatWithSpec, streamChatWithSpec

Constructor Details

VertexAIClient

public VertexAIClient(String model)

VertexAIClient

public VertexAIClient(String model,
 float temperature)

VertexAIClient

public VertexAIClient(String model,
 float temperature,
 Integer maxTokens)

VertexAIClient

public VertexAIClient(String model,
 float temperature,
 Float topP,
 Integer maxTokens)

VertexAIClient

public VertexAIClient(String baseUrl,
 String model,
 float temperature,
 Float topP,
 Integer maxTokens,
 String apiKey,
 String projectId,
 String location)

Method Details

chat

public com.tnsai.llm.ChatResponse chat(String message,
 Optional<String> systemPrompt,
 Optional<List<Map<String,Object>>> history,
 Optional<List<Map<String,Object>>> tools)

streamChat

public Stream<String> streamChat(String message,
 Optional<String> systemPrompt,
 Optional<List<Map<String,Object>>> history,
 Optional<List<Map<String,Object>>> tools)