Google launched its Gemini 3.8 Flash model on July 30, just three weeks after Gemini 3.7 Flash, bringing the total number of Flash models released since May to four. The model excels at coding and matches more expensive models on certain benchmarks while remaining significantly cheaper. The flagship Gemini 3.5 Pro model, which CEO Sundar Pichai had promised for June, remains missing in action: it was still in testing in July and its webpage still shows « coming soon. » On the Artificial Analysis Intelligence Index, Gemini 3.8 Flash now ranks tenth, closely followed by Meta’s new Muse Spark 1.3 model. Google is now betting on recursive self-improvement (RSI) for its Flash models, whose APIs process 22 billion tokens per minute, up from 16 billion the previous quarter.
Source: Read the original article
