Roughly half of its total venture capital funded Arcee AI to train Trinity-Large-Thinking. This 400-billion parameter open reasoning model targets Claude Opus performance in agentic tasks. The high spend highlights the extreme cost of competing with frontier models. Developers now have a massive open-weight alternative for complex, multi-step reasoning workflows.