AI Dictionary › AI Models

Open Weight Models

Modelli Open Weight

Open weight models are AI models whose trained parameters are publicly downloadable: anyone can run them on their own servers, study them, fine-tune them. Examples: Meta's Llama, Mistral, Google's Gemma, DeepSeek. They contrast with closed models (GPT, Claude, Gemini) accessible only via API.

Definition

"Open weight" is not strictly "open source": training data and code often remain private, and licenses impose conditions of use. The distinction matters in debates over transparency and safety.

For companies, the advantages are control and privacy: the model runs on your own infrastructure, data never leaves, costs are predictable and there's no dependency on a single vendor. The costs: you need technical expertise to run them, and the best closed models generally still perform better. The typical choice: open weight for sensitive data and high volumes, closed APIs for the most complex tasks.

Related terms

• Llama
• Mistral
• DeepSeek
• SLM (Small Language Model)
• Fine-tuning

More in AI Models

• ChatGPT
• Gemini
• Copilot
• Mistral
• Llama
• Claude
• GPT (Generative Pre-trained Transformer)
• DeepSeek

Put it into practice

• Real scenario: apply this concept with an instant AI verdict.
• Daily gym: short drills to train dictionary concepts.
• Playoff game: challenge other players, elimination-style.

Say what you mean. Get what you need.

Grace Certified, the AI coach that trains and certifies your prompt engineering, by Agora Intelligence.