Your AI SaaS business, ready by this week!
Try demoYour AI SaaS business, ready by this week!
Try demoAn open-weight model is an AI model whose trained weights are published, so anyone can download it and run it on their own hardware or through any hosting provider.
Closed models, like most GPT, Claude and Gemini versions, are only available through their maker’s API. Open-weight models publish the trained parameters, usually under a licence that sets terms for commercial use.
Families include Llama, Mistral, Qwen, DeepSeek and Gemma. You can run them locally, on a rented GPU server, or through hosts such as OpenRouter and Groq that often price them well below closed models.
"Open weight" is not always "open source": the training data and code may stay private, and licences vary. Check the licence before using a model in a paid product.
Aikeedo can use open-weight models through Ollama on your own server or through hosted providers such as OpenRouter, alongside closed models from OpenAI, Anthropic and Google.
Discover more terms connected to this one
A local LLM is a language model that runs on hardware you control, such as your own server or computer, instead of being called through a cloud provider’s API.
An LLM (large language model) is an AI model trained on very large amounts of text to predict the next token, which lets it write, summarise, translate, reason and hold a conversation.
Fine-tuning is further training an existing model on your own examples so it learns a style, format or narrow task; for adding knowledge, RAG is usually cheaper and easier to update.