TII Releases Falcon H1R 7B Open-Source Reasoning AI Model

The Technology Innovation Institute (TII) of Abu Dhabi released Falcon H1R 7B, an open-source AI reasoning model specialized for coding, mathematics, and agentic tasks. The 7 billion parameter model employs a hybrid architecture that combines Transformer neural networks with Mamba sequence elements to increase throughput and reduce latency.

TII reports that the model matches or outperforms larger systems, including Alibaba's Qwen3 32B and Microsoft's Phi 4 Reasoning Plus 14B, on specialist benchmarks. The model achieved a coding accuracy of 68.6% and scored 88.1% on the AIME-24 mathematics benchmark while requiring significantly less computing power.

Released under the Falcon TII Licence via Hugging Face, the model is designed for developers and public bodies. TII intends for the system to enable advanced reasoning on standard hardware or on-premise infrastructure to reduce operational costs. TII leadership described the model as a leap forward in the reasoning capabilities of compact AI systems, emphasizing its sustainability through low memory and energy use.