Skip to content

Type: AI model

LLaMA-3 is a generation of Meta’s open-weight Llama family, released in sizes including 8B and 70B and serving as a widely adopted baseline for open-model research and fine-tuning. It preceded the 3.1 and 3.3 updates. GROUNDING tracks LLaMA-3’s adoption, derivatives, and benchmark legacy.

Recent Updates

  • 2026-08-12: Cracks in the Foundation: Seemingly Minor Architectural Choices Impact Long Context Extension (cs.CL updates on arXiv.org) · arxiv.orgLong Context Context Engineering Open Source LLMs Olmo · LLaMA · Qwen
  • 2026-09-02: EGT-KG: Evidence-Grounded Typed Knowledge Graph Retrieval for Practical Scientific QA with Small Language Models (breakingnewsofficial) · arxiv.orgRAG
  • 2026-09-02: WebLLM: Browser-Based LLM Inference with GPU Acceleration and OpenAI API Compatibility (breakingnewsofficial) · github.comOpen Source LLMs OpenAI · HuggingFace · LLaMA-2 Hermes-2-Pro-Llama-3 · Phi-3 · Phi-2 Phi 1.5 Gemma-2B · Mistral-7B-v0.3 Hermes-2-Pro-Mistral-7B NeuralHermes-2.5-Mistral-7B OpenHermes-2.5-Mistral-7B Qwen2

FAQ

What is LLaMA-3?

LLaMA-3 is a generation of Meta’s open-weight Llama family, released in sizes including 8B and 70B and serving as a widely adopted baseline for open-model research and fine-tuning. It preceded the 3.1 and 3.3 updates. GROUNDING tracks LLaMA-3’s adoption, derivatives, and benchmark legacy.

What does this page track?

Dated radar mentions, source links, related concepts, and builder-relevant context for LLaMA-3, collected automatically by GROUNDING.

When was LLaMA-3 last mentioned?

LLaMA-3 was most recently mentioned in a radar update dated 2026-09-02.