Ripple-65B Preview Debuts as US Open-Weight SOTA on Agentic Coding at 65B
Azaliamirh · x · 2026-11-17
Ricursive Intelligence released Ripple-65B Preview, an open-weight dense model scoring 80.52 on TerminalBench 2.1 — a new US open-weight SOTA on agentic coding, beating far larger models like Reflection Beam 501B (80.1) and Meta Muse Spark Open 816B (76.4).
- Pretrained on 12T tokens, up to 1M context
- Dense architecture: 300 tok/s code generation, <80ms autocomplete
- Small enough to run locally on NVIDIA DGX Spark
- Available on Hugging Face, Together AI, OpenRouter, Fireworks
It's a general-purpose variant of the model powering the lab's AI chip design platform, claimed to reach frontier open-weight performance at 1/10th the cost. At a hackathon the team showed Ripple building and operating its own code IDE from scratch.
More from coding & agent
- Open-source AI video editor Scarlet ships major update: long-to-shorts, pro NLE handoff — MatthewBerman · 2026-11-17
- JetBrains bundles free Junie Lite with every account, ships baseline model via Junie CLI — SucceededMind · 2026-11-17
- AWS blog: taxonomy of multi-agent design patterns, anti-patterns, and lessons — AWS ML Blog · 2026-11-17
- heybox: Open-Source Touch ID Lock Stops AI Agents From Reading Your HEY Email — seanmcdonaldxyz · 2026-11-17
- GitHub Is Breaking Under Agent-First Workflows — Time for Anthropic to Rebuild It? — signulll · 2026-11-17
- Models earn 9x more from switching ideas than refining, but 58% never find the best basin — AlexGDimakis · 2026-11-17