inclusionAI: Ling 3.0 Tiny

Ling 3.0 Tiny from InclusionAI is an efficient mixture-of-experts model that consistently offers among the most competitive pricing, ranking in the 81st percentile for overall cost effectiveness. It also demonstrates notably high reliability, achieving a 91% success rate across evaluations with minimal technical failures or API disruptions. However, its primary trade-off is speed, as the model tends to have longer response times, placing in the 7th percentile for generation latency. Across benchmark evaluations, the model exhibits its greatest strength in complex reasoning, posting an impressive 90.9% accuracy (61st percentile) on logic, inference, and multi-step problem-solving. In contrast, its performance is more modest in technical and structured domains. It achieved a 79.7% accuracy in coding knowledge (31st percentile) and 57.0% accuracy in instruction following (43rd percentile), showing some limitations when managing layered, high-constraint directives. Overall, Ling 3.0 Tiny stands out as a highly dependable, budget-friendly option with a vast 262K context window, making it well-suited for cost-sensitive reasoning workflows where fast generation speed is not a critical requirement.