Open-source AI
- Definition
- AI models released with open weights and (sometimes) training data, allowing anyone to use, modify, and deploy them. Meta's Llama and Mistral's models lead the open-source wave, competing with closed models from OpenAI and Anthropic.
- Why it matters
- Open-source AI is the most effective competitive pressure in the industry. Every time Meta releases a new Llama model, it resets pricing expectations and forces closed-source providers to justify their premiums. Open-source also provides sovereignty: governments, enterprises, and researchers can run models without sending data to third-party APIs. The strategic dynamics are complex: Meta open-sources models to commoditize the model layer (where its competitors charge) while retaining its advantage in distribution (Facebook, Instagram, WhatsApp). For enterprises, open-source provides insurance against vendor lock-in and enables customization that closed-source APIs cannot match.
- In practice
- Meta's Llama 3 (April 2024) was downloaded millions of times and spawned an ecosystem of fine-tuned variants. Mistral's models, including the MoE-based Mixtral, are used in production by thousands of companies. The open-source inference ecosystem (vLLM, llama.cpp, Ollama) makes it practical to run these models on commodity hardware. Hugging Face hosts 400,000+ open models. The economic impact: companies running open-source models report 60-90% lower inference costs compared to closed-source APIs for equivalent-quality tasks. The trade-off is engineering investment: self-hosting requires infrastructure, monitoring, and security expertise that managed APIs abstract away.
We cover business & strategy every week.
Get the 5 AI stories that matter — free, every Friday.
Related terms
Open weight
A model whose trained parameters are publicly downloadable but whose training data and code may not be shared. Most 'open-source' models are technically open-weight, an important legal and strategic distinction.
Closed-source AI
AI models whose weights, training data, and architecture are proprietary and accessible only through APIs. OpenAI, Anthropic, and Google run closed-source models, monetizing via usage-based pricing.
LLM (Large Language Model)
A neural network trained on massive text corpora to predict and generate language. LLMs like GPT-4, Claude, and Gemini are the foundation of the current AI wave, powering chatbots, coding tools, and enterprise automation.
Foundation model
A large, general-purpose model pre-trained on broad data that can be adapted to many downstream tasks. GPT-4, Claude, Gemini, and Llama are all foundation models. The term signals massive upfront investment and wide applicability.
Know the terms. Know the moves.
Get the 5 AI stories that matter every Friday — free.
Free forever. No spam.