Unsloth
Forecasting & Applied ML Free Trial
Features
- Trains AI models faster with optimized memory usage
- Supports Llama, Mistral and Gemma models
- Supports 4-bit and 16-bit LoRA
- Runs on single or multiple GPUs, with multi-node in Enterprise
- Accelerates inference on NVIDIA, Intel and AMD hardware
- Automates training and memory optimization
- Runs in the browser via Google Colab or Kaggle
- Free open-source version available
Pricing Details
Unsloth offers three main plans: Free, Professional, and Enterprise, with pay-as-you-go options and enterprise services. The free version is open source and supports Llama 1-3, Mistral, and Gemma models, with 4 and 16-bit LoRA and a single GPU, and can be started via Google Colab or Kaggle for free. The Pro plan offers 2.5x faster training, 20% less VRAM consumption, and support for MultiGPU up to 8 units. The Enterprise plan offers up to 30x training acceleration, 30% higher resolution, and full Multi-node support, with customized pricing.