Installation¶

Requirements¶

Python 3.11 or higher
negmas >= 0.11.3
litellm >= 1.80.11

Install from PyPI¶

pip install negmas-llm

Or using uv:

uv add negmas-llm

Install from Source¶

git clone https://github.com/yasserfarouk/negmas-llm.git
cd negmas-llm
pip install -e .

Provider-Specific Setup¶

Cloud Providers¶

Most cloud providers require API keys. Set them as environment variables:

# OpenAI
export OPENAI_API_KEY="your-key"

# Anthropic
export ANTHROPIC_API_KEY="your-key"

# Google Gemini
export GOOGLE_API_KEY="your-key"

# Cohere
export COHERE_API_KEY="your-key"

# Mistral
export MISTRAL_API_KEY="your-key"

# Groq
export GROQ_API_KEY="your-key"

# Together AI
export TOGETHER_API_KEY="your-key"

# OpenRouter
export OPENROUTER_API_KEY="your-key"

# DeepSeek
export DEEPSEEK_API_KEY="your-key"

# Hugging Face
export HF_TOKEN="your-token"

Local Models with Ollama¶

Install Ollama
Pull a model: ollama pull llama3.2
Start the Ollama server (runs by default on http://localhost:11434)

from negmas_llm import OllamaNegotiator

negotiator = OllamaNegotiator(model="llama3.2")

Local Models with vLLM¶

Install vLLM: pip install vllm
Start the server: python -m vllm.entrypoints.openai.api_server --model your-model

from negmas_llm import VLLMNegotiator

negotiator = VLLMNegotiator(
    model="your-model",
    api_base="http://localhost:8000/v1"
)

Local Models with LM Studio¶

Download LM Studio
Load a model and start the local server

from negmas_llm import LMStudioNegotiator

negotiator = LMStudioNegotiator()  # Uses default localhost:1234

Verifying Installation¶

import negmas_llm

# List available negotiator classes
print(negmas_llm.__all__)