Converge Provider

Multi-provider LLM abstraction layer for the Converge runtime.

Website: converge.zone | Docs: docs.rs | Crates.io: converge-provider

Installation

[dependencies]
converge-provider = "0.2"

Related Crates

Crate	Version	Description
converge-core	0.6.1	Runtime engine, agent traits, capabilities
converge-provider	0.2.3	14+ LLM providers, model selection
converge-domain	0.2.3	12 business use cases

Supported Providers

Provider	Models	Region
Anthropic	Claude 3.5 Sonnet, Haiku, Opus 4	US
OpenAI	GPT-4o, GPT-4o-mini, GPT-4 Turbo	US
Google Gemini	Gemini Pro, Flash	US
Alibaba Qwen	Qwen-Max, Qwen-Plus, Qwen3-VL	CN
DeepSeek	DeepSeek Chat, Coder	CN
Mistral	Mistral Large, Medium	EU
xAI Grok	Grok models	US
Perplexity	Online models (web search)	US
OpenRouter	Multi-provider gateway	US
Baidu ERNIE	ERNIE models	CN
Zhipu GLM	GLM-4 models	CN
Kimi (Moonshot)	Moonshot models	CN
Apertus	EU digital sovereignty	EU
Ollama	Local models (Llama, Mistral, etc.)	Local

Features

Model Selection Engine

Cost-aware: VeryLow → VeryHigh cost tiers
Latency-aware: Interactive, Realtime, Batch classes
Quality-scored: 0.0-1.0 quality ratings
Capability matching: Tool use, vision, reasoning, code
Data sovereignty: Region-aware selection (US, EU, CN, Local)

Capability Registry

Unified discovery for LLM, embedding, and reranking providers
Capability-based selection
Local preference support

Vector Stores

In-memory store (testing/development)
LanceDB embedded store (optional feature)
Qdrant distributed store (planned)

Embedding & Reranking

Qwen3-VL multimodal embeddings
Two-stage retrieval pipeline support

Prompt Optimization

EDN format: ~40% token reduction vs markdown
XML wrapping: Claude-specific optimization
Structured response parsing: Extract proposals with confidence

Quick Start

use converge_provider::{AnthropicProvider, LlmProvider, LlmRequest};

// Create provider from environment variable
let provider = AnthropicProvider::from_env("claude-3-5-sonnet-20241022")?;

// Make a request
let request = LlmRequest::new("Analyze market trends for Q4")
    .with_max_tokens(1000)
    .with_temperature(0.7);

let response = provider.complete(&request)?;
println!("Response: {}", response.content);

Model Selection

use converge_provider::{ModelSelector, AgentRequirements, CostClass};

let selector = ModelSelector::default();

// Fast and cheap for simple tasks
let requirements = AgentRequirements::fast_extraction()
    .with_max_cost(CostClass::Low);

let result = selector.select(&requirements)?;
println!("Selected: {} / {}", result.selected.provider, result.selected.model);

Factory Pattern

use converge_provider::factory::{create_provider, can_create_provider};

// Check if provider is available (API key set)
if can_create_provider("anthropic") {
    let provider = create_provider("anthropic", "claude-3-5-sonnet-20241022")?;
    // Use provider...
}

Feature Flags

[dependencies]
converge-provider = { version = "0.2", features = ["lancedb"] }

Feature	Description
`lancedb`	LanceDB embedded vector store
`qdrant`	Qdrant distributed vector store (planned)
`neo4j`	Neo4j graph store (planned)
`all-vector`	All vector stores
`all-stores`	All stores

Environment Variables

Each provider requires an API key environment variable:

Provider	Environment Variable
Anthropic	`ANTHROPIC_API_KEY`
OpenAI	`OPENAI_API_KEY`
Gemini	`GOOGLE_API_KEY`
Qwen	`DASHSCOPE_API_KEY`
DeepSeek	`DEEPSEEK_API_KEY`
Mistral	`MISTRAL_API_KEY`
Grok	`XAI_API_KEY`
Perplexity	`PERPLEXITY_API_KEY`
OpenRouter	`OPENROUTER_API_KEY`

Repository

This crate is part of the Converge project.

Standalone repo: github.com/kpernyer/converge-provider

License

MIT

converge-provider 0.2.3