Together AI Provider

Together AI hosts 200+ open-source models with an OpenAI-compatible API. Great for running large open models without vendor lock-in.

Install

bashpip install "mtpx[togetherai]"

This installs the openai SDK. Alternatively, install the native Together SDK:

bashpip install together

API Key Setup

Option 1: .env file (recommended)

1. Install dotenv support:

bash   pip install python-dotenv

Or with the MTP extra:

bash   pip install "mtpx[dotenv]"

1. Create a .env file in your project root:

text   TOGETHER_API_KEY=your_key_here

1. Load it in your code before creating the provider:

python   from mtp import Agent

   Agent.load_dotenv_if_available()  # reads .env file

Get a free API key at api.together.ai ($1 credit on signup).

Option 2: System environment variable

bash# Linux/macOS
export TOGETHER_API_KEY="..."

# Windows PowerShell
$env:TOGETHER_API_KEY="..."

Quick Start

pythonfrom mtp import Agent
from mtp.providers import TogetherAI

Agent.load_dotenv_if_available()  # loads TOGETHER_API_KEY from .env

provider = TogetherAI(model="meta-llama/Llama-4-Scout-17B-16E-Instruct")
tools = Agent.ToolRegistry()
agent = Agent(provider=provider, tools=tools)

reply = agent.run_loop("What is 25 * 4 + 10?")
print(reply)

Parameters

Capabilities

Recommended Models for Tool Calling

• meta-llama/Llama-4-Scout-17B-16E-Instruct — Best tool use (default)
• meta-llama/Llama-3.3-70B-Instruct-Turbo — Fast, reliable
• Qwen/Qwen2.5-72B-Instruct-Turbo — Excellent reasoning
• deepseek-ai/DeepSeek-V3 — Top-tier reasoning
• mistralai/Mixtral-8x22B-Instruct-v0.1 — Strong + fast

Full Example

pythonfrom mtp import Agent
from mtp.providers import TogetherAI

Agent.load_dotenv_if_available()

provider = TogetherAI(
    model="meta-llama/Llama-4-Scout-17B-16E-Instruct",
    temperature=0.0,
    max_tokens=4096,
)

tools = Agent.ToolRegistry()
agent = Agent(provider=provider, tools=tools, debug_mode=True)

reply = agent.run_loop(
    "Calculate (25 * 4) + 10",
    max_rounds=3,
)
print(reply)

Notes

• Together AI prefers the native together SDK when available, falls back to OpenAI client at https://api.together.xyz/v1.
• Model IDs use the format org/model-name (e.g., meta-llama/Llama-4-Scout-17B-16E-Instruct).
• Widest model selection of any provider — great for comparing model performance on the same task.
• Competitive pricing (~$0.18/1M tokens for 70B models).

Source

src/mtp/providers/together_provider.py