A 20-billion parameter language model from Fireworks AI, optimized for fast inference and cost-effective deployment. At its size, FARE 20B balances general language capability with practical resource efficiency, making it well-suited for high-throughput text generation workloads where frontier-scale models would be overkill.