You can update it manually by downloading Antingravity from the web and installing it (there’s no need to unninstall).
You can also update it via cmd:
winget upgrade Antigravity
1 Like
Spoke too soon. Error 500 came back after the last update. Gemini 3 is not available anymore and 3.1 doesn’t work. I can use just GEmini Flash, Sonnet and Opus.
Trajectory ID: 22a4001a-aa13-4824-a0f4…………..
Error: HTTP 500 Internal Server Error
Sherlog:
TraceID: 0xad15bf………….
Headers: {"Alt-Svc":["h3=\":443\"; ma=2592000,h3-29=\":443\"; ma=2592000"],"Content-Length":["109"],"Content-Type":["text/event-stream"],"Date":["Fri, 20 Feb 2026 14:06:59 GMT"],"Server":["ESF"],"Server-Timing":["gfet4t7; dur=618027"],"Vary":["Origin","X-Origin","Referer"],"X-Cloudaicompanion-Trace-Id":["ad15bf16………"],"X-Content-Type-Options":["nosniff"],"X-Frame-Options":["SAMEORIGIN"],"X-Xss-Protection":["0"]}
{
"error": {
"code": 500,
"message": "Internal error encountered.",
"status": "INTERNAL"
}
}
I get the Agent execution terminated due to error. for weeks now with pro subscription. This is sad.
one of these days I hope to leave this page instead of lingering here checking fruitlessly hoping that someday Google will fix antigravity… … Dr Curtis hoareau leaps from error log to error log … hoping each time will be the next leap that fixes his agent;
‘
Bug Report: Showstopper - Account-Bound Application Lockup and Enterprise Context Collision
1. System Information
- Application: Google Antigravity
- Version: 1.107.0 (Distro: 0c7d350c3a9e8639ea238cc996ec4f6dcf1e35cd)
- Electron Version: 39.2.3
- Node.js Version: 22.20.0
- OS: Microsoft Windows 11 Pro
- OS Version: 22H2
- OS Build: 10.0.22621
2. Problem Description
Google Antigravity fails to initialize correctly and becomes unresponsive (“locks up”) when authenticated with a personal Google account that has a historical “developer status” (e.g., past Google Cloud Project usage). The application’s backend incorrectly propagates an Enterprise/Managed context to the personal account, causing a client-side collision that prevents the internal agent architecture from starting. This is a showstopper bug for affected users, rendering the IDE’s core features unusable.
3. Current Symptoms (As of Feb 21, 2026)
When the application starts with the affected account, the user experiences a “lockup” where AI features are completely non-functional. The logs from today’s session confirm this is due to a cascade of critical initialization failures:
- The core Agent Session service fails to load because it’s treated as an “UNKNOWN service.”
- This causes the Language Server Client to fail its initialization.
- Ultimately, the Extension Host becomes unresponsive, freezing all agent-based functionality.
4. Log-Based Evidence & Root Cause Analysis
The evidence demonstrates a clear chain of failure, beginning with a backend identity conflict and resulting in the client-side lockup detailed above.
A. The Root Cause: Backend Enterprise Conflict
Log analysis confirms the user’s account is being incorrectly assessed, triggering a demand for an Enterprise license that should not apply to a personal account. This is the origin of the entire bug.
2026-01-25 06:36:28.632 [info] E0125 06:36:28.631754 12248 log.go:354] encountered conflicting override on experiment: cascade-knowledge-config2026-01-25 06:36:29.742 [info] E0125 06:36:29.740961 12248 log.go:362] agent executor error: PERMISSION_DENIED (code 403): You are currently configured to use a Google Cloud Project but lack a Gemini Code Assist license. Please contact your administrator to request a license. (#3501)
B. The Result: Client-Side Structural Failure (As of Feb 21, 2026)
This backend conflict directly causes the client’s core services to fail on startup. Today’s logs show that essential agent-related services are not being found or initialized.
2026-02-21 08:19:14.721 [error] [createInstance] fZ depends on UNKNOWN service agentSessions.: Error: [createInstance] fZ depends on UNKNOWN service agentSessions.2026-02-21 08:20:33.874 [error] [Extension Host] Failed to update user status Error: LanguageServerClient must be initialized first!
C. The Symptom: Application Lockup (As of Feb 21, 2026)
The failure of these core services leads to the unresponsive state the user experiences.
2026-02-21 08:19:52.122 [info] Extension host (LocalProcess pid: 5128) is unresponsive.
5. Steps to Reproduce
- Launch Google Antigravity (v1.107.0) on Windows 11 (Build 22621).
- Sign in with a personal Google account that has any prior history of Google Cloud Project creation or API enablement.
- Observe that the Language Server fails to initialize and the Extension Host becomes unresponsive within 60 seconds of startup.
- Verification: Sign out of the account. Observe that the application no longer “locks up” (though AI features remain disabled, as expected), confirming the account is the direct trigger.
6. Requested Resolution
Implement a “Force Personal Tier” toggle in the authentication flow or settings.json. This would instruct the client to ignore backend GCP project discovery and propagation, providing a critical override for users whose personal accounts are permanently flagged with historical developer metadata.
Report generated on February 21, 2026. All data cross-verified against system logs and configuration files for accuracy.”
1 Like
Holy crap!
You just nailed the exact issue I had with a personal account used in the past on google Studio (paid api) and Google Cloud. I thought it was something else but this checks all the marks.
Thank you so much, I will fill a report issue immediately on Antigravity.
F.
Same problem here on Windows, MacOs and Ubuntu. It works on a free account… as there is no support nor information from Google I cancelled my renewal.
…but wait…. it gets better….Im sure someone at google realized that the recent free tier accounts were fine and then decided to say… “hold my prebiotic soda"
check this: “Gemini 3.1 Pro is not available on this version. Please upgrade to the latest version.“
see image attached .. if 3.1 is not the latest model then what is? Where is the latestt model in the list for me to click on, or am I just out of quota cause that mechanism seems to be in place when google wants to complain about my claude usage..? I mean I see no official release but where is the check for updates button recently? Can I opt into a beta release? Will I collect my pension before this issue is resolved? WHy are oral-b tooth brush head replacement so expensive-its just a tiny piece of plastic?
its that matrix game where you have to go into the terminal to try an find out easter egg, the only way I am able to get a handle on whats going on is to actually get Gemini CLI to look into it based on updates and information across the web, why instead of direct access to the the models why dont they allow that intermediary agent that diagnoses and queries google backend to make sure things are correct, you would have some kind of feedback loop that would at least let us know about what the the problem is. This is really just like ridiculous.. i imagine they are handling many different entry points and the complexity has just gotten too much for someone to approve the right logical path to be able to have a user account and correct billing connected up.
Same problem for me yesterday moring, solved downloading the latest version from the website and overwriting my installation. Log out/ log in and done.
It’s silly, I know.
hello @Abhijit_Pramanik same issue for me since the 12/02/2025 : can you help me ?
Trajectory ID: f9d799d4-c
Error: HTTP 500 Internal Server Error
Sherlog:
TraceID: 0xa40424f879c2cee9
Headers: {“Alt-Svc”:[“h3=“:443”; ma=2592000,h3-29=“:443”; ma=2592000”],“Content-Length”:[“109”],“Content-Type”:[“text/event-stream”],“Date”:[“Mon, 23 Feb 2026 15:12:35 GMT”],“Server”:[“ESF”],“Server-Timing”:[“gfet4t7; dur=515”],“Vary”:[“Origin”,“X-Origin”,“Referer”],“X-Cloudaicompanion-Trace-Id”:[“a40424f879c2cee9”],“X-Content-Type-Options”:[“nosniff”],“X-Frame-Options”:[“SAMEORIGIN”],“X-Xss-Protection”:[“0”]}
{
“error”: {
“code”: 500,
“message”: “Internal error encountered.”,
“status”: “INTERNAL”
}
}
The reported HTTP 500 Internal Server Error has been escalated to our engineering team and is currently under investigation.
We appreciate your patience as we work to enhance the Antigravity experience.
1 Like
An update on the current status would really be appreciated.
Do we need to wait hours, days, or weeks for a solution?
For me, the Ulta subscription hasn’t been working for 8 days now.
3 Likes
I am also getting the error :Agent terminated due to error
Trajectory ID: c58754da-249c-44a7-b16b-…
Error: HTTP 400 Bad Request
Sherlog:
TraceID: 0xaf2404e0a36…
Headers: {“Alt-Svc”:[“h3=“:443”; ma=2592000,h3-29=“:443”; ma=2592000”],“Content-Length”:[“282”],“Content-Type”:[“text/event-stream”],“Date”:[“Thu, 26 Feb 2026 13:56:00 GMT”],“Server”:[“ESF”],“Server-Timing”:[“gfet4t7; dur=1107”],“Vary”:[“Origin”,“X-Origin”,“Referer”],“X-Cloudaicompanion-Trace-Id”:[“af2404e0a364c342”],“X-Content-Type-Options”:[“nosniff”],“X-Frame-Options”:[“SAMEORIGIN”],“X-Xss-Protection”:[“0”]}
{
“error”: {
“code”: 400,
“message”: “{“type”:“error”,“error”:{“type”:“invalid_request_error”,“message”:“prompt is too long: 103514 tokens \u003e 102385 maximum”},“request_id”:“req_vrtx_011CYWo6vMAX8oBeQzJ6zKHi”}”,
“status”: “INVALID_ARGUMENT”
}
}
Yes, here too. Same problem, tried everything.
Trajectory ID: 75dc666e-aef6-4fa4-b061-....
Error: HTTP 500 Internal Server Error
Sherlog:
TraceID: 0xf6ba3859731....
Headers: {“Alt-Svc”:[“h3=":443"; ma=2592000,h3-29=":443"; ma=2592000”],“Content-Length”:[“109”],“Content-Type”:[“text/event-stream”],“Date”:[“Thu, 26 Feb 2026 14:14:36 GMT”],“Server”:[“ESF”],“Server-Timing”:[“gfet4t7; dur=354”],“Vary”:[“Origin”,“X-Origin”,“Referer”],“X-Cloudaicompanion-Trace-Id”:[“f6ba38597313cc1f”],“X-Content-Type-Options”:[“nosniff”],“X-Frame-Options”:[“SAMEORIGIN”],“X-Xss-Protection”:[“0”]}
{
“error”: {
“code”: 500,
“message”: “Internal error encountered.”,
“status”: “INTERNAL”
}
}
Well for me it was the age verification. GL everybody
1 Like
@Abhijit_Pramanik
Any updates ? we are paying for nothing while waitng
It is honestly unbelievable. We are looking at more than 10 days of total Antigravity service outage for a AI Ultra user. No feedback, no ETA, and no official status update. As a paying subscriber, I do not feel treated like a professional customer. This is unacceptable.
1 Like
Nothing… not a word from Google, and no way to contact the support. In my case it doesn’t work since 2 week on 3 different computers (Ubuntu, MacOs, and Windows). I tried everything (reset, cleaning, re-downloading, removing extensions and mcp). As we are starting a new month I was hopping some kind of reset… but no, nothing. My subscription expires in a week, I’m interested to see if downgrading to a free account will solve the problem.
Unfortunatly for Google because of their lack of support I’ve subscribed last week to a yearly plan on Claude.
@juergen_Meyer Totally agree… not even a word to say “OK Guys, we confirm the bug and are working on it”. Unacceptable, free plans are working better, so we are really stupid to pay.
Gemini works in VSCode and other IDE plugins using the same Google subscription and account.
Antigravity consistently fails with:
HTTP 500 Internal Server Error
Agent execution terminated due to error
X-Cloudaicompanion-Trace-Id: 8156683b4a31…
Reproduced:
- Multiple Macs
- Fresh installations
- Same account
- Different networks
Colleague in same city with same subscription works fine.
Conclusion:
Gemini inference is functional.
Antigravity Agent execution layer fails at account-level.
Please reset or reinitialize my account binding in the Antigravity Agent backend.
Google is breaking my heart. Pure theft of time to keep checking “if it’s up now” when people started getting comfortable having this a part of their daily workflow. no transparency when the models are down. The Windsurf/Antigravity team goes to sleep with the kind of service outage that is unthinkable from an SLO perspective. This is creating a clown factory.
2 Likes