Tag
Google's Gemini 4 matches GPT 6 Astra on the Artificial Analysis Benchmark while being 40% cheaper, highlighting its cost-performance advantage.
An Epoch AI report indicates that the cost of achieving a specific level of AI performance has dropped sharply, with a 725-fold reduction in inference cost demonstrated by OpenAI o3 and GPT-5.6 Luna on benchmarks like GPQA Diamond.
Bellwethr is developing an open methodology for tracking the real USD cost of a single inference token from capable models, with a draft benchmark suite and community contributions underway.