Image created with Ideogram v3. Image prompt: Late‑90s boy‑band cover “Llama Drama”: five cuddly llamas wearing boy‑band outfits among members; ponchos with rainbow trim; desert sunset backdrop; chrome rodeo‑style logo.
“The Leaderboard Illusion – Identifies systematic issues that have resulted in a distorted playing field of Chatbot Arena – Identifies 27 private LLM variants tested by Meta in the lead-up to the Llama-4 release https://x.com/arankomatsuzaki/status/1917400711882797144
[AINews] Llama 4’s Controversial Weekend Release • Buttondown https://buttondown.com/ainews/archive/ainews-llama-4s-controversial-weekend-release/
“Research reveals gaming of Chatbot Arena : companies test multiple private variants and cherry-pick results while hoarding 63% of community data. https://x.com/fdaudens/status/1917671335758594474
“That’s a wrap on the LlamaCon 2025 keynote! In just over two years, Llama has surpassed 1 billion downloads and established itself as the open ecosystem leader in AI. We’re continuing to support the growth and development of the Llama ecosystem with today’s announcements: The https://x.com/AIatMeta/status/1917278290441822674
“We just announced a major leap forward in AI inference: Groq is partnering with Meta to accelerate the official Llama API giving developers the fastest way to run the latest Llama models with no tradeoffs (starting with Llama 4). What developers get with Groq + Meta: 👉 Speeds https://x.com/JonathanRoss321/status/1917621705503080554
“At LlamaCon 2025, Meta announced: —Standalone Meta AI app with a social ‘discover’ feed to take on ChatGPT —Llama API free preview —Lama Guard 4 (12B), LlamaFirewall, and Prompt Guard —Colab with Groq and Cerebras for faster inference https://x.com/rowancheung/status/1917473779069968505
“Major updates from LlamaCon! We’re advancing AI security with new open-source Llama protection tools and new AI- powered solutions for the defender community. Developers can now access: — Llama Guard 4, a customizable safeguard that supports protections for text and image” / X https://x.com/AIatMeta/status/1917271400118902860
“Meet Solo Tech, one of the 10 international recipients of the second Llama Impact Grants. Solo Tech uses Llama to offer offline, multilingual AI support for underserved rural communities with limited internet access. This grant will help them to equip 50 rural centers with AI https://x.com/AIatMeta/status/1917727629601616030
“Today at LlamaCon, we announced the 10 international recipients of the second Llama Impact Grants! The Llama Impact Grants are aimed at fostering innovation and creating economic opportunities through open-source AI. This year’s recipients showcase a diverse range of solutions, https://x.com/AIatMeta/status/1917274585189568870
““We built this place on open source…” Meta Chief Product Officer Chris Cox took to the stage to kick off LlamaCon 2025, reflecting on our long legacy of open source contributions. 🧵 https://x.com/AIatMeta/status/1917353526088589409
Everything we announced at our first-ever LlamaCon https://ai.meta.com/blog/llamacon-llama-news/
“Qwen3-235B Base seems to be benefiting from its 94 layers compared to Llama-4 Mavericks 48 layers or DeepSeeks 61 layers, which are both much larger models https://x.com/scaling01/status/1916986267700506700
Meta previews an API for its Llama AI models | TechCrunch https://techcrunch.com/2025/04/29/meta-previews-an-api-for-its-llama-ai-models/
“meta.llama4-reasoning-17b-instruct-v1:0 https://x.com/btibor91/status/1917232574344384522
“The Leaderboard Illusion – Identifies systematic issues that have resulted in a distorted playing field of Chatbot Arena – Identifies 27 private LLM variants tested by Meta in the lead-up to the Llama-4 release https://x.com/arankomatsuzaki/status/1917400711882797144?s=46
“It turns out that Meta had 27 different models on LM Arena prior to the launch of Llama 4, but they announced it as if they had one model that topped the leaderboard. An extreme example of benchmark hacking (which other labs also do to lesser degrees). https://x.com/emollick/status/1917435868702257538
