Dense 65B open-weight language model from Ricursive Intelligence, trained for agentic coding with up to 1M context.
Updated EU status checked Last shown
Sources: the vendor's own pages (linked above) and creator videos at the linked timestamps. EU status is information, not legal advice; see how we check.
Ripple-65B Preview is a dense 65-billion-parameter Transformer language model from Ricursive Intelligence, a lab building AI for chip design. It is a general-purpose variant of the proprietary model behind the company's chip design platform. It was pretrained on 12 trillion tokens and supports context lengths up to 1M tokens.
The vendor reports 80.52 on TerminalBench 2.1, which it describes as a U.S. open-weight state of the art for agentic coding. It compares this with larger U.S. open-weight models such as Reflection Beam 501B (80.1) and Meta Muse Spark Open 816B (76.4). These figures are the vendor's own and have not been independently verified here. Ricursive says the full R&D cost was about $32M.
The architecture is a thin-and-deep dense design (d_model 5120, 80 layers) with sliding-window and global attention, gated attention, global NoPE and value residuals. The small size lets it run on local hardware such as NVIDIA DGX Spark. Creators report weights on Hugging Face under Apache 2.0 (about 130 GB in BF16, 57 GB in FP4). Hosted access is listed on Together AI, GMI Cloud, OpenRouter and Fireworks, with one creator citing $0.20/$0.60 per million input/output tokens. This is a preview release, and the model is text and code only.
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
US-made open-weights 65B dense LLM scoring 80.5% on Terminal Bench 2.1 versus other open models. It has a 1M token context, runs locally on a MacBook or DGX Spark, and costs $0.20/$0.60 per million tokens.
Dense 65B open model trained from scratch for ~$32M, claimed state-of-the-art for agentic coding among US open models, with 1M context. Released on Hugging Face under Apache 2 (130 GB BF16, 57 GB FP4).
Closest neighbours by meaning, not just by category.
Model · OpenBMB
Dense 2B-parameter open-weights language model from OpenBMB for on-device and edge deployment, Apache-2.0 licensed.
Model
35B Mixture-of-Experts open-weights model from InternScience for long-horizon agentic search, engineering and tool use.
Model · Poolside AI
Poolside's open-weight Laguna 2.1 coding models: S (118B MoE, 8B active, 1M context) and XS (33B MoE, 3B active).
Model · OpenBMB
1B-parameter dense open-weights language model from OpenBMB for on-device use, with hybrid thinking and 128K context.
Model · Meta
Meta's open-weights 30B dense model for local agents, tuned for tool use and long tasks, Apache 2.0 licensed.
Model · Z.ai
Z.ai's open-weights 753B MoE language model, a post-trained GLM-5.2 aimed at coding, agentic tasks and cybersecurity.