Any model, any provider

Jido does not lock you into a model provider. The req_llm package gives you a single interface to OpenAI, Anthropic, Google, Mistral, and local models. Swap providers by changing configuration, not code.

At a glance

Item	Summary
Best for	Teams evaluating model options, anyone building provider-agnostic agents
Core packages	jido_ai, req_llm, llm_db
Package status	`req_llm` (Stable), `llm_db` (Stable), `jido_ai` (Beta)
Supported providers	OpenAI, Anthropic, Google Gemini, Mistral, local models (Ollama, LM Studio)
Key idea	Your agent code stays the same. The model is configuration, not architecture.

Supported providers

Provider	Adapter	Status
OpenAI	`ReqLLM.Adapters.OpenAI`	Stable
Anthropic	`ReqLLM.Adapters.Anthropic`	Stable
Google Gemini	`ReqLLM.Adapters.Google`	Stable
Mistral	`ReqLLM.Adapters.Mistral`	Stable
Local (Ollama, LM Studio)	`ReqLLM.Adapters.OpenAI` (compatible API)	Stable

All adapters implement the same interface. Your agent code does not change when you switch providers.

Provider-agnostic code with req_llm

req_llm wraps provider differences behind a consistent request/response contract:

# Works with any provider. Just change the model string.
{:ok, response} = ReqLLM.chat(:my_client, [
  %{role: "user", content: "Summarize this support ticket"}
], model: "anthropic/claude-sonnet-4-20250514")

# Same code, different provider
{:ok, response} = ReqLLM.chat(:my_client, [
  %{role: "user", content: "Summarize this support ticket"}
], model: "openai/gpt-4o")

No adapter swapping, no interface changes. The model string determines the provider.

Swap models without changing agent code

Define your agent once. Change the model in configuration:

defmodule MyApp.SupportAgent do
  use Jido.AI.Agent,
    name: "support_agent",
    description: "Customer support agent",
    tools: [MyApp.Tools.OrderLookup],
    system_prompt: "You help customers with order questions."
end

# Production: use Anthropic
config :my_app, MyApp.SupportAgent,
  model: "anthropic/claude-sonnet-4-20250514"

# Development: use a local model
config :my_app, MyApp.SupportAgent,
  model: "ollama/llama3"

Your agent logic, tools, and system prompt stay identical. Only the model changes.

Model capability tracking with llm_db

llm_db maintains a database of model metadata: context windows, pricing, supported features. Use it to make runtime decisions about which model to use:

# Find models that support function calling with at least 100k context
models = LlmDB.list(
  capabilities: [:function_calling],
  min_context: 100_000
)

This is useful when you need to route requests to different models based on task requirements or cost constraints.

What to explore next

Agent foundations:How Jido agents work
Tools:Give agents tools
Coordination:Agents that work together
BEAM advantages:BEAM for AI builders

Get Building

Pick a provider, configure req_llm, and connect it to an agent. Then read Give agents tools to give your agent capabilities beyond conversation.