Latest
  1. Johns Hopkins Bayview tells 3,182 patients an employee put their health data into a public AI tool
  2. China’s AI compute access expected to double in 2027, with US-designed chips about 65 per cent
  3. UK commits £100m to national AI Biosecurity Lab
  4. Moonshot AI releases Kimi K3.1 Pro with 1-million-token context window
  5. Residents in Newport and Cardiff oppose hyperscale datacentres near homes
  6. Study: Grok 4.7 picked Republicans in 92% of forced midterm answers, GPT-6.1 Sol picked Democrats in 97%
  7. Attackers exploiting the PHP flaws hand the break-in to AI agents, Unit 42 says
  8. Elvis Presley’s rightsholders sue Anthropic over training on 274 songs

Latest / Frontier models and safety

Model releaseFrontierUSReported

Meta releases Llama 3.1 405B as open weights, trained on over 16,000 H100 GPUs

See how this story connects

Source: Meta AI blogAlso reported by Maginative, The Stack

Meta released Llama 3.1 on 23 July 2024: a new 405-billion-parameter model and upgraded 8B and 70B versions, all with 128,000-token context and eight languages. The 405B model was trained on over 15 trillion tokens using more than 16,000 NVIDIA H100 GPUs, and Meta called it the first frontier-level open source AI model. Meta said developers could use outputs from Llama models, 405B among them, to improve other models.

Why it matters

Anyone could download and fine-tune a model Meta said rivalled the top closed systems, and use its outputs to train their own smaller models.

When
23 Jul 2024
ID
WTA-N-2024-0043
Who
Meta, NVIDIA

Summary by wheresthe.ai, from Meta AI blog and the reporting credited above. We could not open the primary document, so it is marked Reported. Checked against the source on 9 Oct 2026.

Read the original at Meta AI blog

Source copy: download the page text as saved on 18 Oct 2026 (PDF). Also held by the Internet Archive.

Next stop on this Storyline: Led toMeta releases Llama 4 Scout and Maverick as open-weights multimodal models

Cite this story

WTA-N-2024-0043: “Meta releases Llama 3.1 405B as open weights, trained on over 16,000 H100 GPUs.” wheresthe.ai, 23 Jul 2024. https://wheresthe.ai/d/meta-llama-3-1-405b-open-weights-2024-07/ (source: Meta AI blog).

Spotted an error? Email [email protected].