Meta confirmed it will release open-source versions of its next generation AI models, continuing the strategy the company has followed since shipping the first Llama model in February 2023.
No specific timeline or model names were announced. Meta's pattern has been consistent: release its most capable models under a restricted commercial license - which it calls open source but technically doesn't meet the Open Source Initiative's definition - alongside smaller or more permissive variants. Both tiers typically ship together.
What Three Years of Llama Has Built
Llama 3, released in April 2024, became the most widely deployed open-source model family in the industry. Hundreds of AI tools and commercial products are built on it - from local inference apps to B2B software that uses AI for specific tasks. For developers and smaller companies that can't afford GPT-4-scale API costs for every query, Llama-based models are the practical alternative.
When Meta releases a new generation, it resets the cost floor for everything built on top. Products using open-source model backends get faster and more capable without raising their prices.
The Crowded Open-Source Model Market
This announcement arrives as the field has become genuinely competitive. Google's Gemma series, Microsoft's Phi-3, and Mistral's models all compete for the same use cases Llama once dominated almost alone. Meta's continued commitment to open releases matters not just for its own products but for maintaining competitive pressure on closed-source alternatives.
For anyone using AI tools that run locally or rely on Llama-based backends - tools like LM Studio, Ollama, or various specialized workflow apps - another round of capability upgrades is on the way.