ollama-bench
v1.0.3
Published
A CLI tool to benchmark Ollama models performance
Downloads
244
Readme
Ollama Benchmark Script
A command-line tool to benchmark and compare the performance of Ollama language models. Measures tokens per second, memory usage, and total processing time.
Setup
1. Install Ollama
Choose your platform:
- Windows: Download Installer
- macOS: Download App
- Linux: Run:
curl -fsSL https://ollama.com/install.sh | sh
- Docker: Pull and run:
docker pull ollama/ollama docker run -d -v ollama:/root/.ollama -p 11434:11434 ollama/ollama
2. Start Ollama Server
Before running any benchmarks, make sure the Ollama server is running:
# On Linux/macOS terminal or Windows PowerShell
ollama serve
For Windows users, you can also run Ollama from the system tray after installation.
3. Install Benchmark Tool
Install globally:
npm install -g ollama-benchmark
Or run directly with npx:
npx ollama-benchmark <model1> [model2] [model3]
Usage
# Using global installation
ollama-benchmark smollm:135m qwen2.5:0.5b
# Using npx (no installation required)
npx ollama-benchmark smollm:135m qwen2.5:0.5b
Troubleshooting
If you encounter errors, check:
- Is the Ollama server running? (
ollama serve
) - Can you access
http://localhost:11434
? - Do you have enough RAM for your chosen models?
Available Models
See ollama.com/library for all available models.
License
MIT