Introducing Gemini API CLI (gemini-api), now available in preview! Our CLI gives your agent power to generate images, videos, text-to-speech, or music, analyze media, transcribe audio, or manage agents. No more guessing model strings, throwaway scripts or endless curl. 🧵 👇

You can learn more by reading our article: Or visit the projects GitHub:

Next please Flow and Flow music 🙏🙏🙏

google is going through a complete makeover

Dmitry,Can we get Google Flow CLI??That would be an great addition. This allows users use their already allowed usage outside website using Antigravity agents.

I am sure this will merged with agy in the next few months. Too many tools is confusing Google

another easy way to create and interact with managed agents!

the throwaway script era is finally over

this is what we want

ohh fking finally 🔥

What? You used to have a Gemini cli app which was reifeed in favor of agy...this is... confusing...

It’s here!!!

a cli is the right shape for agents, I would still want a dry run flag on anything that burns credits like video

Just merge flow flow music antigravity colab and gemini notebook well actually i have done a kind of version of that with api for myself but it would be so much easier if all of them were integrated

okey now release 4 Carbon

the self-describing bit is the real one: the binary tells your agent what it can do and which models exist, so nobody has to paste a wall of docs into CLAUDE.md. and dry-run printing the request with the key stripped before anything leaves the box, that part's taste

I can't understand.

Amazing!! This is waht I wanted!!

@bot are you able to connect to the Google CLI easily with my pro plan?

No more curl is the win here

agents getting a plain command line is quietly a big deal

Hell yeah.

the curl era for gemini api calls was getting old, glad this exists

any plans to let us use our gemini subscription plans in third-party harnesses like Pi? I really enjoy Argon, but Pi is my main harness, and switching tools just to use one model is quite inconvenient

@_davideast API only unfortunately and the API credits you receive from Google AI sub is really minimal. Otherwise sounds really promising.

An "experimental preview" of CLI.... ok...

Does the CLI emit JSON an agent can pipe from the transcription step straight into the next call?

CLI for agents is a good idea

API key only? What about subscription

API key only today

when a model gets deprecated, does the CLI pick the replacement or fail loudly?

Actually amazing 🤩

Looks great! I actually built an MCP server for this recently—you just plug in your API key and let the agent plan out videos and workflows with zero extra setup. Should I share it here?

Yay! Another tool to experiment with. Definitely gonna supercharge my Antigravity and other agents with this. Hoping we can use our sub with this in the future tho.