Aleph Alpha

Aleph Alpha's Luminous and Pharia language models — European-built LLMs with strong German and multilingual capabilities.

Install

pip install synapsekit[aleph-alpha]

Setup

export ALEPH_ALPHA_API_KEY=your-api-key

Usage

from synapsekit.llm.aleph_alpha import AlephAlphaLLM
from synapsekit import LLMConfig
import os

config = LLMConfig(
    model="luminous-supreme-control",
    api_key=os.environ["ALEPH_ALPHA_API_KEY"],
    provider="aleph-alpha",
)

llm = AlephAlphaLLM(config)

# Streaming
async for token in llm.stream("Erkläre maschinelles Lernen auf Deutsch"):
    print(token, end="")

# Generate
response = await llm.generate("What are the Luminous model capabilities?")

Available models

Model	Notes
luminous-supreme-control	Flagship, instruction-tuned
luminous-supreme	Highest quality
luminous-extended-control	Balanced, instruction-tuned
luminous-base-control	Fast, lightweight
pharia-1-llm-7b-control	Pharia 7B, instruction-tuned

Auto-detection

The RAG facade auto-detects Aleph Alpha for luminous-* and pharia-* model prefixes:

from synapsekit import RAG

rag = RAG(model="luminous-supreme-control", api_key="...")
rag.add("Your document text here")
answer = rag.ask_sync("Summarize this.")

tip

Luminous models are particularly strong for German-language tasks and EU-based deployments where data residency matters.