LLM benchmarks

Google has launched Gemini 3.7 Flash, cutting per-token pricing by 50% while posting significant improvements in software engineering, web development, and agentic workflow benchmarks. The release, arriving just three weeks after Gemini 3.6 Flash, signals an accelerating competitive cadence at the AI inference layer. ECI Research data shows enterprise AI-assisted coding is scaling fast, making first-pass code accuracy and integration friction the metrics that matter most.