The new AI beast has entered the chat — LLaMA 4.0 is officially here!
AI at Meta just dropped the LLaMA 4 series — three open-source models that are outclassing the likes of GPT-4.5, Claude 3.7, and Mistral 3.1.
Introducing the stars:
a. LLaMA 4 Scout
• 17B active parameters with 16 experts
• 10M token context window (yep, industry-leading!)
• Outperforms Gemma 3, Gemini 2.0 Flash-Lite, and Mistral 3.1 on major benchmarks
b. LLaMA 4 Maverick
• 17B active parameters with 128 experts
• Best-in-class image grounding for real visual understanding
• Beats GPT-4o and Gemini 2.0 Flash across key benchmarks
• Matches DeepSeek v3 in reasoning & coding — with half the parameters
• Scores 1417 ELO on LMArena chat — unmatched performance-to-cost ratio
This isn’t just an upgrade — it’s a game-changer.
Read more here>go.fb.me/gmjohs