Third-party integrations

The Together AI API is OpenAI-compatible, so most third-party SDKs work by pointing them at https://api.together.ai/v1 and supplying your Together API key. The integrations below ship dedicated Together support, with first-class clients, helpers, or providers. For agent frameworks (CrewAI, LangGraph, DSPy, PydanticAI, AutoGen, Agno, Composio), see the dedicated pages under Framework integrations. If you’re migrating from the v1 Python SDK, see the Python SDK v2 migration appendix.

Integration guides

Each of these integrations has a dedicated guide with installation steps and runnable examples:

Hugging Face

Call Together models through Hugging Face Inference clients and the Hub’s provider routing.

Vercel AI SDK

Generate text, stream, call tools, and get structured outputs with the @ai-sdk/togetherai provider.

Mastra

Point Mastra’s model router at Together models to build agents and workflows.

Next.js

Build a streaming chat app with the Together SDK, the Vercel AI SDK, or Mastra.

LangChain

LangChain is a framework for building context-aware, reasoning applications powered by LLMs. The langchain-together package provides chat models and embeddings.

Shell

pip install --upgrade langchain-together

Python

from langchain_together import ChatTogether

chat = ChatTogether(
    model="moonshotai/Kimi-K3",
    extra_body={"reasoning": {"enabled": False}},
)

for chunk in chat.stream("Tell me fun things to do in NYC"):
    print(chunk.content, end="", flush=True)

For more LangChain integration details, see the LangChain provider docs.

LlamaIndex

LlamaIndex is a data framework for connecting custom data sources to LLMs. Together AI works with LlamaIndex through the OpenAILike LLM and dedicated embedding classes.

Shell

pip install llama-index llama-index-llms-openai-like

Python

import os
from llama_index.llms.openai_like import OpenAILike

llm = OpenAILike(
    model="moonshotai/Kimi-K3",
    api_base="https://api.together.ai/v1",
    api_key=os.environ["TOGETHER_API_KEY"],
    is_chat_model=True,
    is_function_calling_model=True,
    temperature=0.1,
)

response = llm.complete("Explain large language models in 500 words.")
print(response)

For more LlamaIndex integration details, see the LlamaIndex Together LLM docs.

LiteLLM

LiteLLM is an open-source gateway with a single OpenAI-format interface for over 100 LLM providers, available as a Python SDK and a proxy server. LiteLLM 1.100.0 and later ships official Together AI support, with a model registry that syncs against the serverless catalog daily.

Shell

pip install "litellm>=1.100.0"
export TOGETHERAI_API_KEY=<your_key>

Python

import litellm

response = litellm.completion(
    model="together_ai/zai-org/GLM-5.3",
    messages=[
        {
            "role": "user",
            "content": "What are some fun things to do in New York?",
        }
    ],
)
print(response.choices[0].message.content)

Streaming (stream=True), function calling (tools), JSON mode (response_format), and reasoning_effort all pass through in the OpenAI format. Reasoning models return their reasoning in a separate reasoning_content field, so content holds only the final answer. To run Claude Code on Together models through the LiteLLM proxy, see the LiteLLM + Claude Code guide.

Helicone

Helicone is an open-source LLM observability platform. Route Together requests through Helicone’s gateway by overriding base_url and adding the auth header.

import os
from together import Together

client = Together(
    api_key=os.environ["TOGETHER_API_KEY"],
    base_url="https://together.hconeai.com/v1",
    default_headers={
        "Helicone-Auth": f"Bearer {os.environ['HELICONE_API_KEY']}",
    },
)

stream = client.chat.completions.create(
    model="Qwen/Qwen3.5-9B",
    messages=[
        {
            "role": "user",
            "content": "What are some fun things to do in New York?",
        }
    ],
    stream=True,
)

for chunk in stream:
    if chunk.choices:
        print(chunk.choices[0].delta.content or "", end="", flush=True)

import Together from "together-ai";

const client = new Together({
  apiKey: process.env.TOGETHER_API_KEY,
  baseURL: "https://together.hconeai.com/v1",
  defaultHeaders: {
    "Helicone-Auth": `Bearer ${process.env.HELICONE_API_KEY}`,
  },
});

const stream = await client.chat.completions.create({
  model: "Qwen/Qwen3.5-9B",
  messages: [
    { role: "user", content: "What are some fun things to do in New York?" },
  ],
  stream: true,
});

for await (const chunk of stream) {
  process.stdout.write(chunk.choices[0]?.delta?.content ?? "");
}

curl https://together.hconeai.com/v1/chat/completions \
  -H "Authorization: Bearer $TOGETHER_API_KEY" \
  -H "Helicone-Auth: Bearer $HELICONE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "Qwen/Qwen3.5-9B",
    "messages": [
      {"role": "user", "content": "What are some fun things to do in New York?"}
    ],
    "stream": true
  }'

Agent frameworks

Each framework below has a dedicated guide with installation, model selection, and runnable examples:

CrewAI: Open-source orchestration for multi-agent workflows.
LangGraph: Stateful, multi-actor applications built on LangChain.
DSPy: Modular AI systems written in code instead of prompt strings.
PydanticAI: Typed agent framework from the Pydantic team.
AutoGen (AG2): Conversational multi-agent systems.
Agno: Open-source library for multimodal agents.
Composio: Tool-use platform for connecting agents to external services.

GET STARTED

SERVERLESS

INFERENCE APIS

DEDICATED MODEL INFERENCE

DEDICATED CONTAINER INFERENCE

GPU CLUSTERS

FINE-TUNING

CODE EXECUTION

ADMINISTRATION

Integration guides

Hugging Face

Vercel AI SDK

Mastra

Next.js

LangChain

LlamaIndex

LiteLLM

Helicone

Agent frameworks

​Integration guides

Hugging Face

Vercel AI SDK

Mastra

Next.js

​LangChain

​LlamaIndex

​LiteLLM

​Helicone

​Agent frameworks

Integration guides

LangChain

LlamaIndex

LiteLLM

Helicone

Agent frameworks