EssentialAI: Rnj 1 Instruct

EssentialAI: Rnj 1 Instruct is an advanced 8-billion-parameter, dense, open-weight model family. Developed by Essential AI and trained from scratch, its core focus lies in programming, mathematics, and scientific reasoning. This model showcases robust performance across a variety of programming languages, supports complex tool-use workflows, and excels in agentic execution environments, including mini-SWE-agent. Designed for versatility, Rnj-1 offers a generous 32K token context window and a maximum output of 4K tokens, making it suitable for intricate tasks. It supports both code generation and streaming capabilities. Pricing is set at an accessible $0.15 per 1M input tokens and $0.15 per 1M output tokens, with free access available through Multi AI.