Providers Overview
Compare the published Namzu provider packages and choose the right integration path for your runtime.
Namzu keeps provider implementations outside @namzu/sdk, but every published provider package plugs into the same ProviderRegistry and LLMProvider contract. The practical question is not how to wire the runtime, but which provider package best matches your deployment reality.
1. Choose a Provider
| Package | Best fit | Notable strength |
|---|---|---|
@namzu/openai | Direct OpenAI usage | Official SDK integration with baseURL overrides |
@namzu/anthropic | Direct Anthropic usage | Native Anthropic Messages API |
@namzu/bedrock | AWS-native deployments | AWS credential-chain support and Bedrock Converse API |
@namzu/openrouter | Multi-vendor model access | One account, many vendors, zero extra runtime dependency |
@namzu/http | Generic compatible endpoints | Zero-dependency fallback for OpenAI- or Anthropic-shaped APIs |
@namzu/ollama | Local Ollama daemon | Local-first open model workflows |
@namzu/lmstudio | Local LM Studio server | Official LM Studio SDK integration |
2. Shared Registration Pattern
Every provider page follows the same runtime shape:
- install
@namzu/sdkand one provider package - call the provider package's
register...()helper once at startup - create the provider through
ProviderRegistry.create({ type: ..., ... }) - pass the returned
providerinto an agent config or callprovider.chat()directly
3. Capability Snapshot
| Package | Tools | Streaming | Notes |
|---|---|---|---|
@namzu/openai | Yes | Yes | Strong default for general-purpose cloud agents |
@namzu/anthropic | Yes | Yes | Anthropic-native behavior |
@namzu/bedrock | Yes | Yes | Best when auth and governance already live in AWS |
@namzu/openrouter | Yes | Yes | Depends on chosen upstream model |
@namzu/http | Yes | Yes | Endpoint must match declared dialect correctly |
@namzu/ollama | Conservative false | Yes | Tool behavior depends on chosen model |
@namzu/lmstudio | Yes | Yes | Loaded model still determines practical quality |
4. Quick Routing Guide
- choose
@namzu/openaiif OpenAI is the primary target - choose
@namzu/anthropicif you want Anthropic-native Messages API semantics - choose
@namzu/bedrockif auth, region, or governance already live inside AWS - choose
@namzu/openrouterif vendor flexibility matters more than direct-vendor coupling - choose
@namzu/httpif the backend is compatible but non-standard for the Namzu package lineup - choose
@namzu/ollamaor@namzu/lmstudiofor local model workflows
5. Recommended Reading Path
| If you need... | Read |
|---|---|
| Help choosing the package in the first place | Provider Selection Guide |
| Registry-level provider wiring | Provider Registry |
Direct chat(), chatStream(), listModels(), and healthCheck() usage | Provider Operations |
| A working first runtime example | SDK Quickstart |
| Detailed package-specific setup | One of the provider pages below |