IBM: Granite 4.1 8B

IBM Granite 4.1 8B is a text-only model with a 131,072-token context window and tool-calling support. It does not support reasoning mode, and structured output support is unconfirmed. The context length is generous for a model at this price tier, making it workable for long-document tasks, but input modalities are limited to text, so multimodal workflows are out. At $0.05 per million input tokens and $0.10 per million output tokens, it sits at the budget end of the market. That pricing may appeal to teams running high-volume text pipelines where cost control matters more than top-tier accuracy. However, its blended benchmark score of 6.5 across only two benchmarks offers limited confidence in how it generalizes; buyers should treat its measured capabilities as preliminary rather than established. It is worth shortlisting for cost-sensitive, text-only use cases, but higher-stakes applications would benefit from a model with broader benchmark coverage.