The Technology Innovation Institute in Abu Dhabi has unveiled Falcon-ASR, a 1.6 billion parameter speech recognition model that achieves state-of-the-art performance on Arabic language benchmarks. The model achieved a 20.92% word error rate across six Arabic test sets, outperforming the previous best published result of 23.17%. On internal evaluations of Emirati dialect speech—a historically underserved use case—Falcon-ASR recorded 22.73% word error rate and 10.19% character error rate, exceeding competing systems including Qwen3-Omni. Developed at TII, the model extends beyond Arabic to support English, French, Spanish, and Portuguese using identical model weights without requiring language flags. Falcon-ASR was trained on diverse Arabic dialects including Emirati and Gulf Arabic, Modern Standard Arabic, and multiple languages, with exposure to real-world recording conditions such as background noise, overlapping speech, music, and telephony effects. The model achieved a mean word error rate of 5.74% on seven English benchmark tests from the Hugging Face Open ASR Leaderboard. The system includes word-level timestamp support for transcriptions, linking each recognized word to its position in the source audio. TII has made the model available through a Hugging Face demo space and indicated that API access and native applications are in development, signaling broader availability for developers and organizations seeking multilingual speech recognition capabilities.