Arcee AI is a U.S.-based open intelligence lab that develops and releases open-weight language models optimized for efficiency and cost-effectiveness. Their flagship Trinity model family (Nano, Mini, Large) is designed for frontier-level performance at a fraction of the inference cost of closed models. Models use online reinforcement learning for continuous post-deployment improvement and are accessible via API, OpenRouter, and direct download. Arcee has raised $24M Series A and partners with Nvidia, Intel, AWS, Microsoft, and Hugging Face. Key features: - Open-weight Trinity model family (Nano, Mini, Large) with Thinking variants - Per-token API pricing significantly below leading closed-model providers - Online reinforcement learning for continuous model improvement after deployment - Available on OpenRouter and via direct API without vendor lock-in - Edge, on-premises, and cloud deployment options for open-weight models - Dedicated coding model (Coder-Large) alongside general-purpose variants
API pricing per model: Trinity Mini $0.045/M input tokens, $0.15/M output tokens; Trinity Large Preview $0.22/M input, $0.85/M output; Coder-Large $0.50/M input, $0.80/M output. Enterprise and on-prem pricing via contact.