Class LlmClient

java.lang.Object
com.codename1.ai.LlmClient
public abstract class LlmClient extends Object

Provider-agnostic chat / embeddings client. Built via one of the static factory methods:

LlmClient gpt    = LlmClient.openai("sk-...");
LlmClient claude = LlmClient.anthropic("sk-ant-...");
LlmClient gemini = LlmClient.gemini("AIza...");
LlmClient ollama = LlmClient.ollama();                   // localhost:11434
LlmClient local  = LlmClient.localOpenAiCompatible(
                       "http://10.0.0.5:8080/v1", "", "qwen2.5-7b");

All calls return AsyncResource so they compose naturally with the rest of the Codename One async API. chatStream additionally fires per-token deltas through a StreamingListener; both that listener and the final AsyncResource complete on the EDT.

Simulator behaviour

When running in the JavaSE simulator with cn1.ai.simulatorRedirect set to ollama (or auto with Ollama detected on the loopback), the static factories transparently route through a local Ollama endpoint instead of the public provider -- so unchanged production code can be debugged offline without API charges.

  • Field Details

  • Constructor Details

    • LlmClient

      protected LlmClient(String baseUrl)

  • Method Details

    • openai

      public static LlmClient openai(String apiKey)
      OpenAI / OpenAI-compatible (Together, Groq, Fireworks, vLLM, Ollama, etc.). Uses the public endpoint by default; override with setBaseUrl(String).

    • anthropic

      public static LlmClient anthropic(String apiKey)

    • gemini

      public static LlmClient gemini(String apiKey)

    • ollama

      public static LlmClient ollama()
      Default Ollama install: http://localhost:11434/v1, model llama3.2.

    • ollama

      public static LlmClient ollama(String defaultModel)

    • ollama

      public static LlmClient ollama(String baseUrl, String defaultModel)

    • localOpenAiCompatible

      public static LlmClient localOpenAiCompatible(String baseUrl, String apiKey, String defaultModel)
      Generic OpenAI-compatible endpoint (llama.cpp server, vLLM, LM Studio, a custom proxy). apiKey may be empty for local services that don't authenticate.

    • chat

      public abstract AsyncResource<ChatResponse> chat(ChatRequest req)
      Non-streaming chat. Equivalent to chatStream with a no-op listener but optimized -- the provider skips the SSE response and returns a single JSON object.

    • chatStream

      public abstract AsyncResource<ChatResponse> chatStream(ChatRequest req, StreamingListener listener)
      Streaming chat. listener fires for every content delta / tool-call fragment on the EDT. The returned AsyncResource completes with the aggregated final response once the stream ends; cancel it to close the underlying socket.

    • embed

      public abstract AsyncResource<EmbeddingResponse> embed(EmbeddingRequest req)

    • getProvider

      public abstract String getProvider()
      One of "openai", "anthropic", "gemini", "ollama", "local". Used by ChatView and tests to vary behaviour by provider.

    • getBaseUrl

      public String getBaseUrl()

    • setBaseUrl

      public void setBaseUrl(String baseUrl)

    • getHttpTimeoutMs

      public int getHttpTimeoutMs()

    • setHttpTimeoutMs

      public void setHttpTimeoutMs(int httpTimeoutMs)

    • runSafetyFilter

      protected String runSafetyFilter(ChatRequest req)
      Helper for subclasses: applies the active SafetyFilter (if any) before the network call. Returns the rejection reason on failure, null to proceed.