Alphabet Inc.’s Google on Wednesday unveiled Gemini CLI, a fully open-source command-line interface that brings its Gemini 2.5 Pro language model directly into developers’ terminals. The software is available immediately in public preview at no charge. Authenticated with a personal Google account, users receive up to 60 model requests a minute and 1,000 a day—allowances Google says are the largest free quota in the market. Gemini CLI supports a one-million-token context window, is released under the Apache 2.0 licence, and can ground answers with real-time Google Search data or be extended through the Model Context Protocol. The move places Google in direct competition with OpenAI’s Codex CLI, Anthropic’s Claude Code and Microsoft-backed GitHub Copilot CLI. The company said the tool shares backend infrastructure with Gemini Code Assist, allowing professional developers to shift to paid Vertex AI or Code Assist plans for higher throughput and enterprise governance while individual users integrate the agent into scripts, CI/CD pipelines or multimodal workflows.
$S | LATEST: @SonicLabs drops Spark, an AI agent that builds Web3 apps—end to end🤖 From DeFi to NFT marketplaces, Spark lets users generate smart contracts, frontends and deploy—all with natural language Sonic says it screens every app for bugs before launch https://t.co/Knbo2K4pgg
A simple Gemini command that changed how I use my smart home https://t.co/TeWt18Qb28
.@SonicLabs has announced Spark, AI platform for creating Web3 Apps https://t.co/uWFUV7ECCF https://t.co/H0iQ9jD9zH