Granite 4.0 H Micro

Granite 4.0 H Micro is IBM's lightweight model in the Granite 4.0 family, positioned for applications requiring fast inference speeds rather than maximum capability. As a micro-tier model, it sits below IBM's more capable Granite variants while offering competitive performance for its size class. The model processes text-only inputs with a 131,000 token context window, providing substantial capacity for document processing and multi-turn conversations. Performance benchmarks show output generation at 464 tokens per second with an 8.7 second time to first token, indicating optimization for throughput over latency. The model handles standard text generation tasks but does not include tool calling capabilities or multimodal processing. Granite 4.0 H Micro targets use cases where speed and efficiency matter more than advanced reasoning capabilities. Its combination of decent context length and fast token generation makes it suitable for high-volume text processing workloads where cost efficiency is prioritized over the sophisticated capabilities found in frontier models.