v2.0.0: Ollama-Only Local LLM Inference

Breaking Changes

• Cloud LLM providers removed — Anthropic, OpenAI, Mistral (cloud), and DeepSeek providers deleted
• No API keys required — QuantCoder now runs exclusively on local models via Ollama
• Task-based model routing — LLMFactory.create(task="coding") auto-selects the right model
• Dependencies removed — openai, anthropic, mistralai packages no longer required

Models

• qwen2.5-coder:32b — code generation, refinement, error fixing
• mistral — reasoning, summarization, chat

Prerequisites

# Install Ollama
curl -fsSL https://ollama.ai/install.sh | sh

# Pull required models
ollama pull qwen2.5-coder:32b
ollama pull mistral

Installation

git clone https://github.com/SL-Mar/quantcoder-cli.git
cd quantcoder-cli
pip install -e .
python -m spacy download en_core_web_sm
quantcoder

What's New in v2.0

• Multi-agent architecture (Coordinator, Alpha, Risk, Universe, Strategy agents)
• Autonomous self-improving pipeline with learning database
• AlphaEvolve-inspired evolution engine
• Library builder for batch strategy generation
• Rich terminal UI with syntax highlighting
• MCP integration for QuantConnect validation and backtesting
• TOML-based configuration

Full Changelog

See CHANGELOG.md