[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"article-5-things-to-know-about-metas-llama-3-rollout-en":3,"article-related-5-things-to-know-about-metas-llama-3-rollout-en":31,"series-industry-bfc20a10-6bc0-42f6-ab76-83ac9846cfcb":84},{"id":4,"slug":5,"title":6,"content":7,"summary":8,"source":9,"source_url":10,"author":11,"image_url":12,"cover_image":12,"category":13,"language":14,"translated_content":11,"related_article_id":15,"keywords":16,"key_takeaways":23,"views":27,"created_at":28,"published_at":29,"topic_cluster_id":30},"bfc20a10-6bc0-42f6-ab76-83ac9846cfcb","5-things-to-know-about-metas-llama-3-rollout-en","5 things to know about Meta’s Llama 3 rollout","\u003Cp data-speakable=\"summary\">\u003Ca href=\"\u002Ftag\u002Fmeta\">Meta\u003C\u002Fa> is expanding Llama 3 access in the US and Europe while limiting some models in the EU.\u003C\u002Fp>\u003Cp>Meta’s Llama 3 rollout gives developers more model choices, but it also shows how open-weight AI runs into regional rules. The family now includes 1 billion, 3 billion, 8 billion, 70 billion, 11 billion, and 90 billion parameter versions.\u003C\u002Fp>\u003Ctable>\u003Cthead>\u003Ctr>\u003Cth>Item\u003C\u002Fth>\u003Cth>Parameter sizes\u003C\u002Fth>\u003Cth>Modality\u003C\u002Fth>\u003Cth>EU availability\u003C\u002Fth>\u003C\u002Ftr>\u003C\u002Fthead>\u003Ctbody>\u003Ctr>\u003Ctd>Llama 3\u003C\u002Ftd>\u003Ctd>8B, 70B\u003C\u002Ftd>\u003Ctd>Text\u003C\u002Ftd>\u003Ctd>Available\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd>Llama 3.1\u003C\u002Ftd>\u003Ctd>Not specified in source\u003C\u002Ftd>\u003Ctd>Text\u003C\u002Ftd>\u003Ctd>Available\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd>Llama 3.2 text\u003C\u002Ftd>\u003Ctd>1B, 3B\u003C\u002Ftd>\u003Ctd>Text\u003C\u002Ftd>\u003Ctd>Available\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd>Llama 3.2 multimodal\u003C\u002Ftd>\u003Ctd>11B, 90B\u003C\u002Ftd>\u003Ctd>Text + image\u003C\u002Ftd>\u003Ctd>Restricted\u003C\u002Ftd>\u003C\u002Ftr>\u003C\u002Ftbody>\u003C\u002Ftable>\u003Ch2>1. Smaller models widen access\u003C\u002Fh2>\u003Cp>The most practical change in the rollout is the addition of smaller text-only models. Meta added 1 billion and 3 billion parameter versions in Llama 3.2, which makes the family easier to use on modest hardware and in local deployments.\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1780772571137-wtvq.png\" alt=\"5 things to know about Meta’s Llama 3 rollout\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>That matters for teams that do not want to depend on large cloud bills or high-end accelerators. Smaller models are better suited to \u003Ca href=\"\u002Fnews\u002Ftether-bitnet-fine-tuning-edge-devices-en\">edge devices\u003C\u002Fa>, internal tools, and lightweight assistants where latency and cost matter more than raw scale.\u003C\u002Fp>\u003Cul>\u003Cli>1B and 3B text-only models\u003C\u002Fli>\u003Cli>Designed for limited compute\u003C\u002Fli>\u003Cli>Useful for on-device or edge use\u003C\u002Fli>\u003Cli>Lower barrier for smaller developers\u003C\u002Fli>\u003C\u002Ful>\u003Ch2>2. Multimodal support expands use cases\u003C\u002Fh2>\u003Cp>Meta also added larger multimodal models in Llama 3.2, with 11 billion and 90 billion parameter versions that can process text and images. That opens the door to image captioning, visual grounding, and document understanding without switching to a separate model family.\u003C\u002Fp>\u003Cp>For developers building product search, support tools, or image-aware workflows, multimodal support is the headline feature. It lets one system handle both language tasks and visual inputs, which can simplify application design.\u003C\u002Fp>\u003Cul>\u003Cli>11B and 90B multimodal models\u003C\u002Fli>\u003Cli>Text plus image input\u003C\u002Fli>\u003Cli>Captioning and visual grounding\u003C\u002Fli>\u003Cli>Document understanding workflows\u003C\u002Fli>\u003C\u002Ful>\u003Ch2>3. Free access is the main adoption play\u003C\u002Fh2>\u003Cp>Meta is offering the Llama 3 family predominantly at no cost, and that is central to the company’s strategy. By lowering access barriers, Meta is trying to pull developers toward its ecosystem instead of closed commercial AI platforms.\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1780772570629-kmtr.png\" alt=\"5 things to know about Meta’s Llama 3 rollout\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>The rollout also runs through major partners such as \u003Ca href=\"https:\u002F\u002Fwww.microsoft.com\">Microsoft\u003C\u002Fa>, \u003Ca href=\"https:\u002F\u002Faws.amazon.com\">Amazon Web Services\u003C\u002Fa>, \u003Ca href=\"https:\u002F\u002Fwww.oracle.com\">Oracle\u003C\u002Fa>, and \u003Ca href=\"https:\u002F\u002Fwww.palantir.com\">Palantir\u003C\u002Fa>. That distribution model makes the models easier to test, deploy, and compare across enterprise environments.\u003C\u002Fp>\u003Cul>\u003Cli>Predominantly free for developers\u003C\u002Fli>\u003Cli>Available through Meta and partner platforms\u003C\u002Fli>\u003Cli>Cloud distribution broadens reach\u003C\u002Fli>\u003Cli>Matches enterprise procurement patterns\u003C\u002Fli>\u003C\u002Ful>\u003Ch2>4. Europe gets a narrower release\u003C\u002Fh2>\u003Cp>Meta expanded access across France, Germany, Italy, Japan, South Korea, NATO, and the European Union, but the EU got a more limited package. The multimodal models are withheld there, while the text-only models remain available.\u003C\u002Fp>\u003Cp>Meta tied that decision to regulatory uncertainty, including GDPR concerns and the EU AI Act. The company also paused plans to train large language models on public adult content from the EU, which shows how regional policy can shape what gets shipped and where.\u003C\u002Fp>\u003Cul>\u003Cli>EU access includes text-only models\u003C\u002Fli>\u003Cli>Multimodal models are restricted in the bloc\u003C\u002Fli>\u003Cli>GDPR and EU AI Act complicate deployment\u003C\u002Fli>\u003Cli>Training plans on EU public content were paused\u003C\u002Fli>\u003C\u002Ful>\u003Ch2>5. Open weights bring both reach and risk\u003C\u002Fh2>\u003Cp>Llama 3 is an open-weight family, which means developers can inspect and adapt it more freely than closed systems. That openness helps adoption, but it also raises concerns about misuse, especially when bad actors can repurpose the models for harmful content or automation.\u003C\u002Fp>\u003Cp>Meta is trying to balance that tension by widening access for developers and public-sector users while limiting some capabilities in stricter regions. The U.S. General Services Administration has even added Llama to its list of sanctioned AI tools for federal agencies, showing how quickly the family has moved into official use.\u003C\u002Fp>\u003Cul>\u003Cli>Open-weight format increases flexibility\u003C\u002Fli>\u003Cli>Misuse risk remains a concern\u003C\u002Fli>\u003Cli>Federal use adds credibility in the US\u003C\u002Fli>\u003Cli>Regional policy shapes model availability\u003C\u002Fli>\u003C\u002Ful>\u003Ch2>How to decide\u003C\u002Fh2>\u003Cp>If you need a small, local model, start with the 1B or 3B text-only options. If your product needs image understanding, the multimodal versions are the better fit, but they are not broadly available in the EU.\u003C\u002Fp>\u003Cp>For teams that want broad deployment and low entry cost, Meta’s partner ecosystem is the easiest path. For regulated buyers, the key question is not just model quality, but where the model can legally run and what kind of data it can touch.\u003C\u002Fp>","5 things to know about Meta’s Llama 3 rollout in the US and EU, including model sizes, regional limits, and developer access.","tech.shepherdgazette.com","https:\u002F\u002Ftech.shepherdgazette.com\u002Fmeta-rolls-out-llama-3-ai-models-us-eu\u002F",null,"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1780772571137-wtvq.png","industry","en","ddac71cf-620a-416f-8b98-f3112d5aeb6f",[17,18,19,20,21,22],"Meta","Llama 3","open-weight AI","multimodal models","EU regulation","developers",[24,25,26],"Llama 3.2 adds smaller text-only models and larger multimodal models.","Meta is expanding access in the US and parts of Europe, but limiting multimodal use in the EU.","Open-weight distribution boosts adoption, while also increasing scrutiny over misuse and regulation.",0,"2026-06-06T19:02:23.052358+00:00","2026-06-06T19:02:23.044+00:00","0aba8265-618f-4651-8595-282f44cbf517",{"tags":32,"relatedLang":43,"relatedPosts":47},[33,35,37,39,41],{"name":20,"slug":34},"multimodal-models",{"name":18,"slug":36},"llama-3",{"name":19,"slug":38},"open-weight-ai",{"name":21,"slug":40},"eu-regulation",{"name":17,"slug":42},"meta",{"id":15,"slug":44,"title":45,"language":46},"5-things-to-know-about-metas-llama-3-rollout-zh","5 個關於 Meta Llama 3 上線的重點","zh",[48,54,60,66,72,78],{"id":49,"slug":50,"title":51,"cover_image":52,"image_url":52,"created_at":53,"category":13},"a71ad261-e32d-44b8-ab3e-4bff5ac98055","denver-hailstorm-roads-damage-checklist-en","Denver hailstorm turns roads into a damage checklist","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1780780693560-vyq8.png","2026-06-06T21:17:48.840853+00:00",{"id":55,"slug":56,"title":57,"cover_image":58,"image_url":58,"created_at":59,"category":13},"2d54715c-e070-4c66-a8c1-cdc114a37eeb","aj-brown-trade-talks-tilt-toward-eagles-en","A.J. Brown Trade Talks Tilt Toward Eagles","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1780779772944-avtt.png","2026-06-06T21:02:25.120935+00:00",{"id":61,"slug":62,"title":63,"cover_image":64,"image_url":64,"created_at":65,"category":13},"18772850-3016-421b-9758-095468cb982c","5-steps-connect-codex-with-deepseek-en","5 steps to connect Codex with DeepSeek","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1780778866728-mnvb.png","2026-06-06T20:47:20.631837+00:00",{"id":67,"slug":68,"title":69,"cover_image":70,"image_url":70,"created_at":71,"category":13},"b7998c1b-8e10-4f65-aef3-59a428f36541","how-to-run-gemma-4-locally-unsloth-en","How to Run Gemma 4 Locally","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1780777065791-3eje.png","2026-06-06T20:17:21.697706+00:00",{"id":73,"slug":74,"title":75,"cover_image":76,"image_url":76,"created_at":77,"category":13},"2982171e-d440-41cf-b188-65cdda134338","vibe-coding-enterprise-software-change-management-en","Vibe coding is changing enterprise software work","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1780774386263-f0ln.png","2026-06-06T19:32:33.912058+00:00",{"id":79,"slug":80,"title":81,"cover_image":82,"image_url":82,"created_at":83,"category":13},"bcb80586-e23e-4822-b98e-d92b65870928","5-reasons-to-use-kimi-k2-5-on-cloudflare-en","5 reasons to use Kimi K2.5 on Cloudflare","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1780769864414-tqp9.png","2026-06-06T18:17:17.730869+00:00",[85,90,95,100,105,110,115,120,125,130],{"id":86,"slug":87,"title":88,"created_at":89},"d35a1bd9-e709-412e-a2df-392df1dc572a","ai-impact-2026-developments-market-en","AI's Impact in 2026: Key Developments and Market Shifts","2026-03-25T16:20:33.205823+00:00",{"id":91,"slug":92,"title":93,"created_at":94},"5ed27921-5fd6-492e-8c59-78393bf37710","trumps-ai-legislative-framework-en","Trump's AI Legislative Framework: What's Inside?","2026-03-25T16:22:20.005325+00:00",{"id":96,"slug":97,"title":98,"created_at":99},"e454a642-f03c-4794-b185-5f651aebbaca","nvidia-gtc-2026-key-highlights-innovations-en","NVIDIA GTC 2026: Key Highlights and Innovations","2026-03-25T16:22:47.882615+00:00",{"id":101,"slug":102,"title":103,"created_at":104},"0ebb5b16-774a-4922-945d-5f2ce1df5a6d","claude-usage-diversifies-learning-curves-en","Claude Usage Diversifies, Learning Curves Emerge","2026-03-25T16:25:50.770376+00:00",{"id":106,"slug":107,"title":108,"created_at":109},"69934e86-2fc5-4280-8223-7b917a48ace8","openclaw-ai-commoditization-concerns-en","OpenClaw's Rise Raises Concerns of AI Model Commoditization","2026-03-25T16:26:30.582047+00:00",{"id":111,"slug":112,"title":113,"created_at":114},"b4b2575b-2ac8-46b2-b90e-ab1d7c060797","google-gemini-ai-rollout-2026-en","Google's Gemini AI Rollout Extended to 2026","2026-03-25T16:28:14.808842+00:00",{"id":116,"slug":117,"title":118,"created_at":119},"6e18bc65-42ae-4ad0-b564-67d7f66b979e","meta-llama4-fabricated-results-scandal-en","Meta's Llama 4 Scandal: Fabricated AI Test Results Unveiled","2026-03-25T16:29:15.482836+00:00",{"id":121,"slug":122,"title":123,"created_at":124},"bf888e9d-08be-4f47-996c-7b24b5ab3500","accenture-mistral-ai-deployment-en","Accenture and Mistral AI Team Up for AI Deployment","2026-03-25T16:31:01.894655+00:00",{"id":126,"slug":127,"title":128,"created_at":129},"5382b536-fad2-49c6-ac85-9eb2bae49f35","mistral-ai-high-stakes-2026-en","Mistral AI: Facing High Stakes in 2026","2026-03-25T16:31:39.941974+00:00",{"id":131,"slug":132,"title":133,"created_at":134},"9da3d2d6-b669-4971-ba1d-17fdb3548ed5","cursors-meteoric-rise-pressures-en","Cursor's Meteoric Rise Faces Industry Pressures","2026-03-25T16:32:21.899217+00:00"]