Gemini 2.5 Pro Tops Coding Benchmark: Beats Claude and GPT-5 to Become the Strongest Code Model

According to the latest data from Google DeepMind, Gemini 2.5 Pro achieved a score of 72.3% on the SWE-bench Verified programming benchmark, surpassing Claude Opus 4.5 and GPT-5 to become the most powerful code generation model. Google also announced a 30% price reduction for Gemini 2.5 Pro on Google AI Studio and Vertex AI to attract more developers to migrate.