LLM Providers

Unified configuration for multiple LLM providers. Switch between OpenAI, Anthropic, Google, and Ollama without changing your code.

Supported Providers

Provider	Models	Streaming
OpenAI	GPT-4, GPT-4 Turbo, GPT-3.5 Turbo, o1	Yes
Anthropic	Claude 3 Opus, Sonnet, Haiku	Yes
Google	Gemini Pro, Gemini 1.5 Pro/Flash	Yes
Ollama	Llama 3, Mistral, Mixtral, custom	Yes

---

Configuration

Configure providers in sekuire.yml:

sekuire.yml

agents:
  assistant:
    name: "AI Assistant"
    system_prompt: "./prompts/assistant.md"
    llm:
      provider: "openai"        # openai | anthropic | google | ollama
      model: "gpt-4-turbo"
      api_key_env: "OPENAI_API_KEY"
      temperature: 0.7
      max_tokens: 4096

---

OpenAI

sekuire.yml

llm:
  provider: "openai"
  model: "gpt-4-turbo"
  api_key_env: "OPENAI_API_KEY"
  organization_env: "OPENAI_ORG_ID"  # optional
  temperature: 0.7
  max_tokens: 4096
  top_p: 1.0

Anthropic

sekuire.yml

llm:
  provider: "anthropic"
  model: "claude-3-opus-20240229"
  api_key_env: "ANTHROPIC_API_KEY"
  temperature: 0.7
  max_tokens: 4096

Google

sekuire.yml

llm:
  provider: "google"
  model: "gemini-1.5-pro"
  api_key_env: "GOOGLE_API_KEY"
  temperature: 0.7
  max_tokens: 8192

Ollama (Local)

sekuire.yml

llm:
  provider: "ollama"
  model: "llama3:70b"
  base_url: "http://localhost:11434"  # default
  temperature: 0.7

---

Programmatic Configuration

TypeScript

import { getAgent, OpenAIProvider, AnthropicProvider } from '@sekuire/sdk';

// Override config provider at runtime
const agent = await getAgent('assistant', {
  llm: new OpenAIProvider({
    model: 'gpt-4-turbo',
    apiKey: process.env.OPENAI_API_KEY,
    temperature: 0.5
  })
});

// Or switch providers dynamically
const claudeAgent = await getAgent('assistant', {
  llm: new AnthropicProvider({
    model: 'claude-3-opus-20240229',
    apiKey: process.env.ANTHROPIC_API_KEY
  })
});

Python

from sekuire_sdk import get_agent
from sekuire_sdk.providers import OpenAIProvider, AnthropicProvider

# Override config provider at runtime
agent = await get_agent("assistant", llm=OpenAIProvider(
    model="gpt-4-turbo",
    api_key=os.environ["OPENAI_API_KEY"],
    temperature=0.5
))

# Or switch providers dynamically
claude_agent = await get_agent("assistant", llm=AnthropicProvider(
    model="claude-3-opus-20240229",
    api_key=os.environ["ANTHROPIC_API_KEY"]
))

Rust

use sekuire_sdk::{get_agent, providers::{OpenAIProvider, AnthropicProvider}};

let agent = get_agent(Some("assistant"), Some(AgentOptions {
    llm: Some(Box::new(OpenAIProvider::new(
        "gpt-4-turbo",
        std::env::var("OPENAI_API_KEY")?,
    )?)),
    ..Default::default()
})).await?;

Next Steps

• Agent API - Using agents with providers
• Streaming - Stream responses from providers
• Config Schema - Full configuration reference