Google appears to have finally gotten into its stride of creating — and then productionizing — some impressive AI releases.
Cursor AI has integrated Google’s Gemini-2.5-Pro model in its IDE. “Gemini 2.5 Pro is available to all Cursor users! You can enable the full 1M context window if you’d like. We’re curious to hear how you think it compares to Sonnet,” Cursor posted on X.
Users are already able to select the Gemini-2.5-Pro experimental model to help with their code. Anthropic’s Claude 3.7-Sonnet was thought by many to be the most capable coding model, but Google’s 2.5-Pro experimental, released last week, had zoomed to the top of most benchmarks.
On the Aider Polyglot benchmark, Gemini-2.5-Pro had managed a score of 74 percent, ahead of Claude Sonnet 3.7’s 64.9 percent and OpenAI’s o3-mini high’s 60 percent. On non-coding benchmarks too, Gemini 2.5-Pro had done particularly well — it had scored the highest ever score of 18.8 percent on the Humanity’s last exam, compared to the 14 percent that o3-mini had managed. In the science focused GPQA diamond benchmark, it scored 84 percent, higher than the 79.7 percent by OpenAI’s o-3 mini. In math’s AIME 2025 benchmark too, Gemini 2.5 Pro was marginally ahead of o3-mini high.
Google had appeared to have been caught napping in the AI race when OpenAI had released ChatGPT in November 2022. Google had invented the transformer, which was the critical breakthrough that led to the AI revolution, but was largely seen to be lagging behind upstarts like OpenAI and Anthropic in creating models and releasing AI products. But over the last couple of years, Google has not only improved its models — Gemini-Pro-2.5 beats most other models on benchmarks — but also seems to be actively deploying them into usable products. Cursor is the biggest player right now in the ‘vibe coding’ space, and Google integrating its top model within days of its release into Cursor shows that the company is finally up and running in the AI race.