Granite 4.0 H 1B
IBM · 2025-10-28 · 1.5B parameters
Granite 4.0 H 1B is IBM's instruction-tuned, text-only 1,461,538,368-parameter Nano model, fine-tuned from Granite 4.0 H 1B Base with supervised fine-tuning, reinforcement learning, and model merging. It is a dense hybrid decoder architecture with 36 Mamba-2 layers, four grouped-query-attention layers, and a 131,072-token context window; IBM positions it for edge, on-device, and latency-sensitive use cases. Its weights are released under Apache 2.0, and it belongs to the Granite language-model family covered by IBM's ISO/IEC 42001 certification.