Unified LLM framework supporting reasoning models, VLM-OCR, GraphRAG, and agentic workflows — 8 providers, 80% test coverage
| LangChain | LlamaIndex | beanllm | |
|---|---|---|---|
| Architecture | Flat chain | Index-centric | Clean Architecture (Facade → Handler → Service → Domain) |
| Reasoning Models | Manual config | Manual config | thinking_budget native support |
| VLM-OCR | Not supported | Not supported | 11 engines + Qwen3-VL / GLM-OCR / DeepSeek-VL2 |
| GraphRAG | Plugin | Plugin | KnowledgeGraph Facade built-in |
| Test Coverage | — | — | 80% (6,340 tests) |
| ORPO Fine-tuning | Not supported | Not supported | Native support (50% less memory than DPO) |
| Providers | OpenAI-heavy | OpenAI-heavy | 8 providers including Grok/xAI |
beanllm is the AI infrastructure layer for careerOS, a production career platform (Spring Boot 3.3, 14 domains, 415 tests):
| Use Case | Model | Volume |
|---|---|---|
| Resume structured extraction | gpt-4o-mini | Per upload |
| Job posting normalization | gpt-4o-mini | Batch, 7 collectors |
| Advisor skill-gap report | gpt-4o | On-demand |
| Match narrative generation | gpt-4o | Daily digest |
| Candidate graph embeddings | text-embedding-3-small (1536-dim) | Per graph rebuild |
careerOS routes all AI calls through beanllm's unified interface. Setting app.ai.provider: mock swaps in stub responses — enabling all 415 tests to run without any live API calls.
pip install beanllm[openai] # or [anthropic], [gemini], [all]import asyncio
from beanllm import Client
async def main():
client = Client(model="gpt-4o")
response = await client.chat(
messages=[{"role": "user", "content": "Explain quantum computing"}]
)
print(response.content)
asyncio.run(main())from beanllm import RAGChain
rag = RAGChain.from_documents("docs/")
result = await rag.query("What is this about?", include_sources=True)| Provider | Install | Env var |
|---|---|---|
| OpenAI | beanllm[openai] |
OPENAI_API_KEY |
| Claude (Anthropic) | beanllm[anthropic] |
ANTHROPIC_API_KEY |
| Gemini (Google) | beanllm[gemini] |
GEMINI_API_KEY |
| Grok / xAI | beanllm[all] |
XAI_API_KEY |
| DeepSeek | beanllm[all] |
DEEPSEEK_API_KEY |
| Perplexity | beanllm[all] |
PERPLEXITY_API_KEY |
| Ollama (local) | beanllm (built-in) |
OLLAMA_HOST (optional) |
| HuggingFace | beanllm[all] |
HUGGINGFACE_API_KEY |
Provider auto-selection, CircuitBreaker, and rate limiting: wiki/providers.md
| Module | Highlights |
|---|---|
| LLM Providers | 8 providers, unified interface, ModelParameterStrategy auto-adapts per-model params |
| Reasoning Models | Native thinking_budget for Claude/OpenAI o-series; <thinking> token filtering |
| RAG Pipeline | Document loaders, 8 vector stores, hybrid search, HyDE, MultiQuery, rerankers |
| GraphRAG | KnowledgeGraph Facade — NER, relation extraction, Neo4j, relationship-aware retrieval |
| VLM-OCR | 11 engines (PaddleOCR, Qwen3-VL, GLM-OCR, DeepSeek-VL2); 3-layer PDF processing |
| Fine-tuning | LoRA/QLoRA, DPO, ORPO (50% less memory vs DPO), KTO |
| Multi-Agent | Sequential, parallel, hierarchical, debate patterns; DAG graph workflows |
| MCP Server | Model Context Protocol server for tool integration |
| Architecture | Clean Architecture layers, request flow diagram |
| Providers | Setup, CircuitBreaker state machine, rate limiting |
| Facade API | Client, RAGChain, Agent usage guide |
| API Reference | Full endpoint specs (client, agent, models, RAG) |
| Decision Log | 3 architectural decisions |
| Playbooks | CircuitBreaker / rate limit incident runbooks |