AI modelsNovember 04, 2026
-28
🧊
65B Model Crushes 975B Giants
A new open-weight model called Ripple-65B reportedly reaches state-of-the-art results among US open-weight models on agentic coding, trained on a budget of roughly $32 million.
#open-weight#agentic coding#LLM#Hugging Face#benchmarks
🔥 What happened
Ricursive Intelligence dropped Ripple-65B Preview, an open-weight model that sets a new U.S. state-of-the-art on agentic coding — beating much larger models like Thinking Machines Inkling (975B) and Meta Muse Spark Open (816B).
💡 Why it matters
The entire R&D cost was $32M — one-tenth of the previous open-weight SOTA. It also needed 70x fewer RL rollouts than Reflection Beam. Efficiency, not raw size, is the new lever.
⚡ Our take
The 'bigger is better' dogma just took a hit. Anyone still brute-forcing MoE scaling should study this dense architecture closely — it might be the blueprint for the next wave.
The title, summary and analysis of this item were produced automatically by an AI system and have not been editorially reviewed. They may contain errors, bias or omissions — when in doubt, read the linked original source.
