#Ollama Provider

Run open-source models locally with Ollama for privacy and cost savings.

#Prerequisites

1. Install Ollama: ollama.ai
2. Pull a model: ollama pull llama3.2
3. Verify: ollama list

#Configuration

[auth]
provider = "ollama"

[model]
provider_id = "ollama"
model = "llama3.2"
base_url = "http://localhost:11434/v1"

#Available Models

Pull models:

ollama pull llama3.2
ollama pull mistral
ollama pull codellama

#CLI Usage

# Use Llama 3.2
savfox -m ollama:llama3.2 exec "Task"

# Use with OSS flag
savfox --oss exec "Task"

#Configuration Options

[model.ollama]
base_url = "http://localhost:11434/v1"
temperature = 0.7
num_ctx = 4096
num_gpu = 1

#GPU Acceleration

Ollama automatically uses GPU when available:

• macOS: Metal (Apple Silicon)
• Linux: CUDA (NVIDIA)
• Windows: CUDA (NVIDIA)

#Memory Requirements

#Troubleshooting

#Connection refused

1. Ensure Ollama is running: ollama serve
2. Check the base URL
3. Verify the port (default: 11434)

#Out of memory

1. Use a smaller model
2. Reduce num_ctx
3. Close other applications

#Slow responses

1. Enable GPU acceleration
2. Use a smaller model
3. Reduce context length