Meta releases LLaMA, sparking the open-weights LLM wave
Confirmed
Importance: historic (5 of 5)The takeaway
Meta released LLaMA (7B–65B) to researchers; LLaMA-13B outperformed GPT-3 on most benchmarks, and after the weights leaked in early March the model seeded a vast open-source ecosystem (Alpaca, Vicuna, llama.cpp).
Status
- Claim
Confirmed
- Our reporting
- High confidence
- Importance
- Historic (5 of 5)
- Last verified
- 29 September 2026
Your AI and this story
- GPT-6 AstraIn its training data
- Claude Opus 5.5In its training data
- Gemini 4 HydrogenIn its training data
- Grok 4.7In its training data
It happened before the cutoffs of all four assistants, so it can be in their training data.
Key facts
- Announced 24 February 2023; paper arXiv 2302.13971
- Sizes: 7B, 13B, 33B, 65B parameters
- Trained only on publicly available data, up to 1.4T tokens
- LLaMA-13B outperformed GPT-3 (175B) on most benchmarks reported
- Weights leaked publicly within about a week
What happened
Meta published a family of Chinchilla-style efficient foundation models under a research license.
Why it matters
Kick-started the open-weights LLM movement that later produced Llama 2/3, Mistral, Qwen and DeepSeek.
Sources
2 sources from 2 sites. Numbers match the chips in the text.
2 sources: 2 primary
Primary
- Introducing LLaMA (Meta AI)ai.meta.com, official
- LLaMA: Open and Efficient Foundation Language Models (arXiv)arxiv.org, paper
Changes
- Filed