Ripple-65B Preview delivers strong agentic coding results as a fully Apache 2.0 open US model

What it does

Ripple-65B Preview is a large language model with 65 billion parameters released by Ricursive Intelligence. It aims to compete as a fully open U.S.-based model under the Apache 2.0 license. The model targets coding tasks and scores highest among open-source American models on the TerminalBench 2.1 coding benchmark, demonstrating strong agentic coding capabilities. The total investment behind Ripple-65B Preview’s development is reported at $32 million.

Why it matters

For developers and businesses seeking an open-weight U.S. alternative to proprietary models, Ripple-65B Preview offers a rare option with transparent licensing and solid coding performance. Its Apache 2.0 license lowers friction for integration and commercial use compared to more restrictive models. However, while it leads in agentic coding among U.S. open-source competitors, Ripple falls short of outperforming other recent western open models like Inkling 2 Small on benchmarks beyond coding, signaling that it is not a universal solution yet. The heavy R&D spend demonstrates serious commitment to building competitive large models outside of traditional Silicon Valley giants but also sets expectations for high-quality and ongoing improvements.

Who it is for

Builders and product teams focused on coding automation will find Ripple-65B’s strengths appealing. Enterprises and startups requiring a model with clear legal usage rights from a U.S. vendor may prefer Ripple over opaque or foreign models. This could reduce legal and compliance risks for sensitive applications. Researchers wanting an open, well-documented large model with substantial performance can leverage Ripple especially in agentic coding tasks. However, those needing more well-rounded or cutting-edge multi-modal capabilities should watch alternatives like Inkling 2 Small.

The catch

Ripple-65B Preview’s edge is narrowly focused on coding benchmarks tracked by TerminalBench 2.1. Its weaker showings on other tests suggest limitations in general natural language understanding or generation. This matters if your use case extends beyond code generation or includes conversation, reasoning, or creative writing. The $32 million R&D tag also signals a capital-intensive project, which may influence pricing or ongoing development speed. Being fully open source under Apache 2.0 is a strength but means community uptake and ecosystem building remain key to its long-term viability.

What to watch next

Observe how Ripple-65B’s performance evolves through updates and how Ricursive Intelligence expands its ecosystem and commercial partnerships. Monitor adoption rates in developer communities and enterprise deals to assess practical traction against competitors like Inkling 2 Small. Watch for improvements addressing Ripple’s current weaknesses on non-coding tasks. Finally, tracking licensing models and cost structures will be important for operators deciding between open U.S.-based solutions and proprietary alternatives.

AI Quick Briefs Editorial Desk