Nous Research, an open-source AI startup backed by crypto venture firm Paradigm, has released a new competitive programming model called NousCoder-14B that matches or exceeds several larger proprietary systems. The model was trained in just four days using 48 Nvidia B200 graphics processors and achieves a 67.87 percent accuracy rate on the LiveCodeBench v6 evaluation.
The release comes at a charged moment, with Anthropic's Claude Code dominating social media discussion since New Year's Day. Nous Research is betting that open-source alternatives trained on verifiable problems can close the gap, and transparency in how these models are built matters as much as raw capability.
NousCoder-14B was built by researcher Joe Li using the company's Atropos framework, which enables any researcher with sufficient compute to reproduce or extend the work. The model relies on "verifiable rewards" - a system where the model generates code solutions, those solutions are executed against test cases, and the model receives a simple binary signal: correct or incorrect.
The training process used dynamic sampling policy optimization (DAPO), which discards training examples where the model either solves all attempts or fails all attempts. The researchers also adopted iterative context extension, first training the model with a 32,000-token context window before expanding to 40,000 tokens.
Nous Research has carved out a distinctive position in the AI landscape as a company committed to open-source releases that compete with proprietary alternatives. The company raised $65 million in a round led by Paradigm and has cultivated a community around its anime-style branding and aesthetic.
