Unsloth adds AMD GPU support for faster local LLM training

Unsloth has released an update that significantly enhances support for AMD GPUs, enabling local LLM training and inference across Windows, WSL, and Linux. This new version offers up to 2x faster performance and requires 70% less VRAM compared to previous configurations, with no loss in accuracy. The update also improves compatibility with AMD's MI300X and MI325X GPUs, supports GGUF and Safetensors formats, and introduces features for running larger models by utilizing system memory and splitting models across multiple GPUs. AI

IMPACT Accelerates local LLM deployment on AMD hardware, potentially lowering barriers for users without NVIDIA GPUs.

RANK_REASON This is a software update for a tool that enhances LLM performance on specific hardware, not a core AI model release or research paper.

Read on Unsloth — Releases →

• AMD
• Blackwell
• Cloudflare
• Gemma 4
• GGUF
• GLM 4.7 Flash
• Hugging Face
• Inkling
• Instinct
• MI300X
• MI325X
• NVIDIA
• Qwen 3.5
• Qwen3.6
• Radeon
• Rocm
• Ryzen
• safetensors
• Unsloth

• infra
• product

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →