Gemini 3 Pro gives Cline more headroom for long-running coding
Blog post from Cline
Gemini 3 Pro, Google's latest flagship model, excels in reasoning, multimodal tasks, and coding, marking a significant advancement from its predecessor, Gemini 2.5 Pro. It achieves high scores on public leaderboards for web development and reasoning benchmarks, while also demonstrating strong performance in terminal and tool-use evaluations. The model is adept at managing a codebase, tools, and artifacts, going beyond merely answering isolated questions. In the Cline environment, Gemini 3 Pro shows a remarkable ability to maintain structure over long tasks and effectively utilize its 1M-token context window, making it more coherent and context-aware than previous iterations. The model's multimodal understanding allows it to integrate code, design, and documentation seamlessly, facilitating more productive workflows. Gemini 3 Pro fits into Cline's structured workflow of project investigation, planning, and execution, raising the ceiling of achievable tasks by keeping earlier context in view and producing comprehensive plans. It is available through various providers, including Cline, OpenRouter, Vercel, Google Gemini, and GCP Vertex, and continues to expand the scope of tasks that can be effectively handed off to an AI agent.