Google releases Gemini API CLI as open-source preview
The dependency-free binary for macOS, Linux and Windows lets AI agents generate media and transcribe audio from the terminal.
Google released an experimental preview of Gemini API CLI, an open-source command-line tool that lets developers and AI agents use Gemini API features directly from the terminal, according to aitimes.com. The announcement was made on October 22, 2026.
The tool ships as a dependency-free binary for macOS, Linux and Windows, and can also be installed from source. It supports generating images, videos, music and speech, analyzing media and transcribing audio through simple commands.
Media generation runs on specific models: image generation uses Nano Banana 2.1, video generation uses Gemini Omni 1.1 Flash, music generation uses Lyria 3.5 and speech synthesis uses Gemini 3.8 Flash TTS. Generated images and videos are written to files, and the command output displays the file paths.
Built for agents and scripted work
Transcription can produce timestamped SRT subtitle files, and YouTube video analysis is also supported. Agents capable of running shell commands can use the tool; Google cites Claude Code, Codex, OpenClaw, Hermes and Cursor as supported examples. Multi-step research or coding tasks can be handed off to background Managed Agents such as Deep Research.
The command set can be inspected offline, with options including --schema, --dry-run, JSON/TOON output and --jq filtering. Google says the tool already works in its Antigravity environment, with native integration planned for later.
Gemini API CLI is open source under the Apache 2.0 license and is currently listed on GitHub at experimental preview version 0.4.x. In October 2026, Google described its AI development ecosystem as having three layers: the Gemini API, the Google Generative AI SDK for Python, TypeScript, Go and Java, and the new command-line tool.
Quick answers
What is Gemini API CLI?
It is Google's open-source command-line tool for using Gemini API features, such as image, video, music and speech generation and audio transcription, from the terminal.
Which platforms does it support?
It is distributed as a dependency-free binary for macOS, Linux and Windows and can also be installed from source.
What version is it at?
Google lists it on GitHub as an experimental preview at version 0.4.x, under the Apache 2.0 license.