langchain-kyma

Kyma API integration for LangChain.

Kyma API is a free LLM gateway providing access to 21+ open-source models (DeepSeek, Qwen, Llama, Gemma, Mistral, and more) via a single OpenAI-compatible endpoint.

Installation

pip install -U langchain-kyma

Setup

Get a free API key at kymaapi.com, then set it as an environment variable:

export KYMA_API_KEY="ky-your-api-key-here"

Usage

from langchain_kyma import ChatKyma

llm = ChatKyma(model="qwen-3.6-plus")

messages = [
    ("system", "You are a helpful assistant."),
    ("human", "What is the capital of France?"),
]
ai_msg = llm.invoke(messages)
print(ai_msg.content)

Model aliases

Use convenient aliases to select the best model for your use case:

ChatKyma(model="best")       # highest quality (qwen-3.6-plus)
ChatKyma(model="fast")       # lowest latency (llama-3.3-70b)
ChatKyma(model="code")       # coding tasks (kimi-k2.5)
ChatKyma(model="reasoning")  # complex reasoning (deepseek-r1)
ChatKyma(model="agent")      # agentic tasks (kimi-k2.5)
ChatKyma(model="vision")     # multimodal (gemma-4-31b)

Streaming

for chunk in llm.stream(messages):
    print(chunk.content, end="", flush=True)

Tool calling

from pydantic import BaseModel, Field

class GetWeather(BaseModel):
    """Get the current weather in a given location."""
    location: str = Field(..., description="City and state, e.g. San Francisco, CA")

llm_with_tools = llm.bind_tools([GetWeather])
ai_msg = llm_with_tools.invoke("What's the weather in Paris?")
print(ai_msg.tool_calls)

Structured output

from pydantic import BaseModel, Field
from typing import Optional

class Joke(BaseModel):
    """Joke to tell user."""
    setup: str = Field(description="The setup of the joke")
    punchline: str = Field(description="The punchline to the joke")
    rating: Optional[int] = Field(default=None, description="Funniness from 1 to 10")

structured_llm = llm.with_structured_output(Joke)
joke = structured_llm.invoke("Tell me a joke about AI")
print(joke)

Async

import asyncio

async def main():
    ai_msg = await llm.ainvoke(messages)
    print(ai_msg.content)

asyncio.run(main())

Available models

Browse all available models at kymaapi.com/dashboard/models or via the API:

import httpx, os

resp = httpx.get(
    "https://kymaapi.com/v1/models",
    headers={"Authorization": f"Bearer {os.environ['KYMA_API_KEY']}"},
)
for model in resp.json()["data"]:
    print(model["id"])

Links

• Kyma API docs
• GitHub
• PyPI