FR version is available. Content is displayed in original English for accuracy.
Advertisement
Advertisement
⚡ Community Insights
Discussion Sentiment
54% Positive
Analyzed from 4920 words in the discussion.
Trending Topics
#google#gemini#cli#antigravity#product#using#claude#code#more#products
Discussion (195 Comments)Read Original on HackerNews
Even if there are competing implementations, in terms of brand recognition, I feel like “Gemini” is more closely associated with Google than “Antigravity”. Why pick the more obscure option?!
Perhaps they felt the sentiment on Gemini CLI was beyond repair, but surely there must be some voice on the inside saying “developers will never adopt our products if we keep killing them”.
And tbh I can’t really argue with that.
The Antigravity harness is by far better than the gemini-cli one. Antigravity also offers models other than Gemini as well. When you say Antigravity, you think of a platform whereas when you say Gemini you think of the model
It's great that gemini-cli is open-source, but that also comes with a bunch of ai-generated issues and pull-requests, which is sure to impede development
Absolutely not? When you say "Antigravity" then the first thing that comes to mind is "yet another IDE" and I have no desire in switching my IDE.
I the other reasons you mentioned could be solved while keeping the Gemini name, but this is a fair point. I didn't realize they offered 3rd party models!
> When you say Antigravity, you think of a platform whereas when you say Gemini you think of the model
Yea I guess if their goal long-term is to be something more akin to Cursor that makes sense, but Anthropic seems to be doing just fine using "Claude" in their naming scheme.
No, I don't.
I've never heard of Antigravity, but have (rarely) used Gemini, so there's that.
In a way, it is exciting to me that people exist that think like this. It is so different than how I think, we could be from different planets.
Gemini was on life support on my side. I barely get to use it due to its subpar performance in coding, which is to be honest the only use I have of it.
And now I read that they spent 4 to 5 months testing 3.5 internally. Let that sink in. By the time they release the world has moved on. I don’t know who makes decisions at Google regarding AI but it saddens me to see this happening. Google should be up there leading but they are lagging against everybody.
How can I justify dropping 100$ per month, for a coding agent that is half a year behind, knowing that Codex or Kimi is going to do much better?
Stock might be ripping but that’s about it.
I wonder though if Google isn't so worried about the viability of their coding AIs and have a longer term view than simply providing coding aids. This might also be indicated by their recent $40B investment in Anthropic, https://www.cnbc.com/2026/04/24/google-to-invest-up-to-40-bi...
...Only time will tell!
They use Gemini for personal assistent to all of their Gmail and co users/customers. They have Google Docs, they have GCP were gemini should support you too.
They also have a lot more languages to support too.
They optimize Gemini for A LOT more than 'just' coding. So its probably a balance act for them. And because they are that rich and have no issues on compute and brain, they can play the long game easily.
If they push their tpu further and continue their build out, they will be able to start training high quality topic optimized models in parallel while everyone else needs the same amount to just train one main model.
Its much better branding than calling every single product "Co-Pilot (tm)".
Now Meets Chat is Hangouts again!
Having a single perfect product strategy with non-overlapping product categories and understandable names is hard for any organization, particularly in a rapidly evolving space.
Its obviously an issue to have multiple mature products be chaotically names.
At this moment antigravity and gemini cli and are hardly mature. Isn't now the perfect time to consolidate?
As if I needed another reason to hate them, they turned our Nest back to shitty thermostats last year by dropping older models from their Google Home service. There's no justification for it other than some product owner wanted to.
We all want this to be the case but it's never the case. It never stops to amuse me how developers of the world fall into the Google trap again and again and again despite knowing better.
Personally have been hurt a lot by the abandonment of Polymer and since then it would not occur to me to touch any Google development product because what's the point really?
How their target audience feels isn’t separate from “analysis” - it’s the input.
EDIT: ... also that the analysis at the "top" is mostly being made by people with the wrong incentives and motives, too.
Antigravity CLI is not - the repo has a README and an animated gif demo: https://github.com/google-antigravity/antigravity-cli
[0] https://xcancel.com/i/status/2056801834652594521
If anything, I suspect closing the source for their coding agent may have been part of the goal.
[1] https://killedbygoogle.com/
Of course, something could’ve been improved, but that’s just how they operate.
I could be completely wrong though
One could argue coding is only a use case and that their models are still killing it overall. However agents are strategic across the board and coding agents are at the forefront. They’ve already lead to new products like CoWork and it’s easy to understand why Google should be doing everything possible to catch up.
Surprised they’re not trying to entice developers away with more heavily subsidized subscription plans. Maybe it’s because as some say those days are ending and soon we’ll all be paying per token. Or maybe it would just put too much of a strain on available compute.
The problem is with your perception of reality. Google doesn't operate for the outside, you're on the outside, Google operates for Google and people in Google care about themselves first, then Google, and then -- if t all, outside.
But that i pay for some 2tb storage and i'm a 'pro' user while not really a 'pro'user and that there is another 'pro' package makes all of that very weird. This is something they need to clean up
I can’t believe our Google account setup is different from any other startup in SF. Anyone have success with this? Do they even have a bot at Google that tracks this attrition?
But they happily take my money for a couple of Workspce accounts.
I've been waiting for the "Google decides kids shouldn't vibe code" headline over the Antigravity age verification shenanigans.
And clicking on "Explore Google AI plans" takes me to...I kid you not...the storage settings page of google drive.
Genuinely can't tell wtf google even wants me to use. Vertex? Gemini? Antigravity? Antigravity 2? Agent platform? Google One? Gemini Enterprise? Google AI?
Don't they have a senior management team that can impose some coordination?
OpenAI came up with GPT 4o, o4, 4.5, 4.1 (which came out later than 4.5 and had a completely different purpose), Microsoft just calls everything a copilot (Github Copilot, Azure Copilot, Microsoft Copilot - all from the same company, completely different things), and Google apparently just picks random words from the comments.
Google has such poor UX flows at times, it doesn’t surprise me.
They do have a senior management team, as can be ascertained by the size of their compensation package.
For the other part, which is actually doing actual management: nope.
- A Chrome DevTools Protocol / Playwright client.
- macOS Seatbelt sandbox (--sandbox flag) with some special Node / v8 stuff.
- Sentry for crash reporting and Unleash for feature flags.
- A SKILL.md system mirroring Anthropic's skills convention.
- Subagents, an artifacts review workflow (slash commands), and conversation rewind.
- Telemetry redaction in several places (good?)
- go-git bundled in there.
- go-enry / linguist's entire language table: many file extension/syntax tags (Cairo, Stacks Clarity, Modelica, KiCad, etc.) bundled in there.
All in all, a 140 MB Go binary with its own browser control stack, sandbox, Git, language detector, skills runtime, and subagent system.
I'm good, I'll stick with pi and codex. Less is more my friends.
> - Subagents, an artifacts review workflow (slash commands), and conversation rewind.
Antigravity CLI, like Gemini CLI, is a copy of most of Claude Code. At least in Antigravity CLI they copied the better UI as well. The scope of copying includes support for definitions of Agents, Skills, Commands, Plugins, MCP and so on. In fact, for some time, the Gemini CLI "extensions" documentation referred directly to Claude Code marketplace repositories. An artifact of this is that for example CLAUDE_PROJECT_DIR is made available to hooks, by Gemini CLI.
If the agent won't tell you what it's programming is, it's not your agent.
Two fast reflections:
1. I personally really doubt you can stay competitive selling such low-agency products to agentic developers, who are used to having access to/ability to see & reform their agentic worlds.
2. Also impressed by the hubris of giving everyone a single month to make the transition! I'd love the Google Graveyard to keep track of how long between announcement and shut down products got; I expect Gemini CLI getting axed for Antigravity CLI with one month transition is close to a record.
I guess it's time to move to something else with a business model that might survive a reorg.
Update: unsubscribed from Google Ultra, moving to Claude Max
It was a complete buggy mess - at one point I asked Gemini why it could not use the network despite having network access enabled in the sandbox settings, and it told me that although it had network access, it couldn't use mdnsresponder while running with the built-in sandbox. Like, how well thought out, network access without DNS.
After burning through about 80% of my 5-hour window of credits, I finally just went sandboxless to get the thing running. It hit the limit pretty quickly. I waited until the 5 hour limit was up, and found the 5 hour window had morphed into a one week window, still drained of credits.
I thought at least I can keep on using Gemini CLI until Google figures out this Antigravity thing. Oh well.
What I've done instead is built a script to create a disposable virtual machine (using incus to manage it).
And then I just run the CLI inside the virtual machine and delete the vm at the end of each day.
Local VMs are heavyweight but useful if you are sandboxing an entire IDE/GUI app like Cursor. With containers it's somewhat annoying to share local files - Distrobox helps with GUI apps and mounting the home directory but loses sandboxing. I have been curious about Flatpak/bubblewrap, but haven't had time to try it.
For now I've settled on containers, but I would like to shift to a remote VM like I have at work.
I dug into it a little bit to see about improving things there, but decided to write a minimal version that better suited my needs instead.
It would be much better if they just gave up on Gemini for coding and exclusively adopted Claude models. Even Deep Mind folks themselves prefer Claude over Gemini[1].
[1]: https://www.businessinsider.com/google-deepmind-ai-tool-divi...
"need to install a complete desktop app to get access to our new CLI"
This is a cluster across the entire product line
gemini-cli had it's own quota, antigravity had it's own quota, and ai studio had it's own free tier quota and I managed to make use of all of them super cheaply.
Now they're finally unifying everything and cutting down, which is less of a cognitive load to keep track of quotas but also fewer benefits
If your organization uses Gemini CLI or our IDE extensions via a Gemini Code Assist Standard or Enterprise license, or if your organization uses Gemini Code Assist for GitHub through Google Cloud, your access remains unchanged. We’ll continue to support Gemini CLI and Gemini Code Assist with access to the latest Gemini models and other updates.
Oh, at least they didn't drop off Enterprise users. I think the general transition is towards building specialized products on top of agents. A lot of people are using claude code, codex, and other subsidized coding agents for non coding purposes as well.
It was included in my employers workspace subscription so I tried it out last june, and that's how I finally understood the power of AI.
Then they announced that it was no longer included in our license and I bought my own Claude license instead, the employer went with another AI company.
So your loss Google.
https://antigravity.google/product/antigravity-cli
They just revamped Gemini CLI. Plus it gets the harness of Antigravity, seems like a straight up upgrade to me?
> run CLI agent with an initial prompt
> tell the agent it isn't allowed to directly reply to the user and must use your tool instead. also all of the CLI's original interactive tools are blocked and it has to use your alternatives
> when the agent uses tools in the MCP, it redirects to your GUI's prompt editor
One may argue that Google’s Antigravity is clunky or cluttered or something worse, but that’s confusing organizational capability with principles.
"One may argue that Google’s Antigravity is clunky or cluttered or something worse, but that’s confusing organizational capability with principles"
The amount of free and biased good will Google gets here in HN is weird.
Now they have Antigravity IDE, Antigravity 2, and Antigravity CLI.
Antigravity locks your chats locally behind .pb files.
Nothing to export your very own data.
OpenAI is best at personal data export. Claude has something at least, despite being quirky. Yet, Google looks very purpose-built to not give anything back.
This is not good for open-source. Claude is not open-source, copilot-cli is not and antigravity-cli isn't either.
Apparently the major players decide to keep the secret agent source, well, secret.
Not that it will be missed much. Using it was the worst experience out of any harness.
That is one reason I avoid Flutter at all costs despite other reasons.
[1]: https://killedbygoogle.com/
What makes it "incomprehensibly bad." in your opinion?
What do the Antigravity quotas mean per plan?
Anyway, one more @ Google Graveyard: https://killedbygoogle.com/
Migration is half assed, lots of extensions and mcps doesnt work Themes are fucked up (why not just copy everything over from geminicli?) agy cli doesn't know about itself and can't comment on basic things about its config ...
#CLI
https://www.joelonsoftware.com/2000/04/06/things-you-should-...
First prompt thought for about 30 seconds, and after the fifth or sixth tool call:
"Our servers are experiencing high traffic right now, please try again in a minute."
Sigh...
which is mainly this part on googles thinking:
> Drop whatever you are doing because it’s not important. What is important is OUR time. It’s costing us time and money to support our shit, and we’re tired of it, so we’re not going to support it anymore. So drop your fucking plans and go start digging through our shitty documentation, begging for scraps on forums, and oh by the way, our new shit is COMPLETELY different from the old shit, because well, we fucked that design up pretty bad, heh, but hey, that’s YOUR problem, not our problem.