Provider Index¶

Amplifier doesn't lock you into a single AI vendor. Every model interaction flows through a provider module — a thin adapter that translates Amplifier's internal request format into whatever a particular LLM API expects. Swap providers, mix providers, or run models locally. The rest of the system never notices.

This page maps out every provider available today, shows you how to configure each one, and explains how multi-provider routing works.

Section Contents¶

The Provider Contract¶

Every provider module implements one function:

complete(request) → response

That's the entire contract. The request carries the conversation history, model parameters, and tool definitions. The response returns the model's output in Amplifier's normalized format. Streaming, tool calls, token counting — all handled inside the provider so nothing upstream needs to care which model answered.

┌──────────────┐     ┌──────────────┐     ┌──────────────┐
│  Amplifier   │────▶│   Provider   │────▶│   LLM API    │
│   Kernel     │◀────│   Module     │◀────│  (any vendor) │
└──────────────┘     └──────────────┘     └──────────────┘

Because every provider speaks the same internal protocol, switching from Claude to GPT is a config change — not a code change.

Official Providers¶

These providers ship with Amplifier and are actively maintained.

Anthropic (Claude)¶

The default provider. Claude models excel at code generation, long-context reasoning, and tool use.

Set up Claude as your provider

# amplifier.yaml
provider:
  name: anthropic
  api_key: ${ANTHROPIC_API_KEY}
  model: claude-sonnet-4-20250514

[Tool: bash] export ANTHROPIC_API_KEY="sk-ant-..."

OpenAI (GPT)¶

Full support for GPT-4o, GPT-4 Turbo, and o-series reasoning models.

Configure OpenAI

provider:
  name: openai
  api_key: ${OPENAI_API_KEY}
  model: gpt-4o

[Tool: bash] export OPENAI_API_KEY="sk-..."

Google Gemini¶

Access to Gemini Pro and Ultra models with native multimodal support.

Configure Gemini

provider:
  name: google
  api_key: ${GOOGLE_API_KEY}
  model: gemini-2.5-pro

Azure OpenAI¶

For teams that need Azure's compliance, data residency, and enterprise controls.

Configure Azure OpenAI

provider:
  name: azure-openai
  api_key: ${AZURE_OPENAI_API_KEY}
  endpoint: https://your-resource.openai.azure.com
  deployment: gpt-4o-deployment
  api_version: "2024-06-01"

Azure requires both an endpoint and a deployment name — the model is tied to the deployment, not specified directly.

Ollama (Local Models)¶

Run models entirely on your machine. No API keys, no network calls, no data leaving your laptop.

Configure Ollama for local inference

provider:
  name: ollama
  endpoint: http://localhost:11434
  model: llama3.1:70b

[Tool: bash] ollama pull llama3.1:70b
[Tool: bash] ollama serve

Ollama is ideal for air-gapped environments, sensitive codebases, or just experimenting without burning API credits.

Community Providers¶

These providers are maintained by the community. They follow the same complete(request) → response contract and install as standard modules.

Community providers are versioned independently. Check each module's docs for supported models and configuration options.

Multi-Provider Configuration¶

Here's where it gets interesting. Amplifier's routing matrix lets you assign different providers to different roles — so your coding tasks use one model while your research tasks use another.

Set up a multi-provider routing matrix

# amplifier.yaml
providers:
  anthropic:
    api_key: ${ANTHROPIC_API_KEY}
  openai:
    api_key: ${OPENAI_API_KEY}
  ollama:
    endpoint: http://localhost:11434

routing:
  coding: anthropic/claude-sonnet-4-20250514
  fast: openai/gpt-4o-mini
  research: anthropic/claude-sonnet-4-20250514
  reasoning: anthropic/claude-sonnet-4-20250514
  creative: openai/gpt-4o
  security-audit: anthropic/claude-sonnet-4-20250514
  vision: openai/gpt-4o

Each role maps to a provider/model pair. When Amplifier dispatches a task, the routing matrix picks the right model automatically. You never manually select — you just work, and the system routes.

Check which provider handles each role

[Tool: mode] operation: current
Active routing matrix: balanced
  coding      → anthropic/claude-sonnet-4-20250514
  fast        → openai/gpt-4o-mini
  research    → anthropic/claude-sonnet-4-20250514
  creative    → openai/gpt-4o

Provider Selection Criteria¶

Choosing providers is a tradeoff. Here's a framework:

Most teams start with a single provider and add more as specific needs emerge. The routing matrix makes it painless to experiment — change one line, see if the results improve.

Quick Setup Checklist¶

For any provider, setup follows the same pattern:

1. Get credentials — API key, endpoint, or local server
2. Set environment variable — export PROVIDER_API_KEY="..."
3. Add to config — provider block in amplifier.yaml
4. Verify — run a simple prompt and confirm the response

Verify your provider is working

> What model are you?

I'm Claude, made by Anthropic. Your Amplifier session is routing
through the anthropic provider with claude-sonnet-4-20250514.

If you see a response, you're connected. If not, check that your API key is set and the endpoint is reachable.

Next Steps¶

With providers configured, explore Routing to understand how Amplifier selects models for different tasks. For building your own provider module, see Custom Tool — the module contract is the same pattern.