XIO

xWire

Posts from every feed QUEST has found across the web — blogs, news, podcasts, channels and more.

Google AI Developers Forum - Latest posts

Latest posts

Firebase Authentication – Token Service API Disabled and Missing IAM Permissions in AI Studio Starter Tier

Hello Google AI Studio Team, I would like to follow up on this issue. After further investigation, it appears that my account does not have the IAM permissions required to enable securetoken.googleapis.com. I would strongly prefer to preserve my existing Firebase project and remain on the Starter Tier if possible. Could a Google AI Studio engineer or platform administrator please: 1. Verify the current status of securetoken.googleapis.com in project 852275327899. 2. Check whether the project…

Has anyone gotten implicit caching to work?

Hi Lucia, thanks for the earlier replies in this thread. We see a related but different symptom on Gemini 3.5 Flash-Lite and 3.1 Flash-Lite (not 3.1 Pro), and would like to understand whether it is intended before we restructure our prompts. You mentioned that caching improvements were released for newer models and are being expanded to older ones. We see a behavioural difference between two models that may be related. Setup: generateContent via the Gemini Developer API (not Vertex), Python…

Firebase Authentication – Token Service API Disabled and Missing IAM Permissions in AI Studio Starter Tier

Google buffed their entire auth ecosystem a few weeks ago, enforcing stricter management on users, specially with Firebase Admin SDK APIs and keys. People forget all the time the “zero trust” thinking when dealing with sensitive stuff… backdoors open everywhere. The error you’re getting says its clearly: The token service API was never used (now is disabled by default) or has been disabled on purpose (either by you or automatically by Google on safety grounds). You need to enable it and give…

Gemini-3.5-transcribe drops the last 2–3 words when audio ends without trailing silence

Model: gemini-3.5-transcribe (Interactions API, REST, audio via Files API) Modes: verbatim and smart (both affected identically) Problem For longer audio files (tested: 241 s and 260 s, German speech, MP3 mono 64 kbit/s 44.1 kHz, no decode errors) whose speech runs right up to the end of the file, the transcript is missing the final 2–3 words – the last sentence is cut mid-way. Both modes cut at exactly the same word. gemini-2.5-flash transcribes the same files completely. What I checked…

Interactions File Search: 3 stores trigger “Max 5 corpora”; GenerateContent works

We’re seeing a reproducible difference between Interactions and GenerateContent using gemini-3.8-flash. Request Result Interactions with one or two real, indexed stores Succeeds Interactions with three distinct stores HTTP 400 GenerateContent with the same three stores and metadata filter Succeeds Streaming GenerateContent with those three stores Succeeds, with expected document citations The Interactions error is: FileSearch.file_search_store_names: [FIELD_INVALID] Max 5 corpora can be…

[Regression] Sandbox mode and Custom Security Presets missing after upgrading from 2.19.1 to 2.28.1 (Windows 10)

Hi everyone, I encountered an issue after updating Antigravity from 2.19.1 to 2.28.1 on Windows 10: the ability to configure custom security settings and run terminal commands in the sandbox seems to have completely disappeared. Environment Antigravity Version: 2.28.1 (Regression observed when upgrading from 2.19.1) OS: Windows 10 (Version 22H2) Previous Behavior (v2.19.1) In version 2.19.1, navigating to Settings → General showed a Security Preset dropdown with the following options: Default…

Request to add Syria to Google Antigravity supported regions following sanctions removal

Hello everyone, I’d like to add my support to this request. Developers exist all over the world, and all of them need access to their tools, especially now that agentic workflows with AI have become part of everyday software development. Google set out with the mission of making knowledge accessible to everyone. The next step is making AI systems accessible too. Anyone working on software should be able to use these tools, regardless of where they live. I respectfully ask the Antigravity team…

[BUG] Google AI Pro — Claude/GPT Weekly Quota at 0% Despite No Usage for Over 7 Days

Hi @sk_Eun, Thank you for the detailed report and for verifying the /usage numbers across both the Antigravity IDE and the Antigravity CLI (Gemini Weekly: 97.36%, Gemini Five Hour: 94.46%, Claude/GPT Weekly: 0.00%). Because the same 0.00% balance and reset timer appear in both surfaces after more than seven days without Claude or GPT usage, this points to a backend weekly quota bucket rollover issue rather than a local display cache problem. We have escalated your report to our engineering team…

No response after 4 weeks - appeal submitted Sep 27, and the case ID does not appear to exist

Hi @or_T , Thank you for following up. While we cannot look up individual private account or appeal records here on the public forum, we want to give you a direct and transparent explanation of how each part of this works so you are not left guessing: How Official Appeal Form Outcomes Work (No Email or Status Portal) The Antigravity Terms of Service Appeal Form does not generate a tracking ID, send confirmation or decision emails, or provide an online status page. When an eligible appeal is…

Feature Request: Option to auto-sync / write generated artifacts directly to the local file system

Hi team, With the recent changes, generated artifacts seem to be held primarily in an ephemeral session/preview state rather than being immediately written to the local project workspace. While keeping things virtual prevents accidental file clutter, it introduces significant friction for developers who rely on: External IDEs and editors that watch the local directory for live file changes. Local linters, test runners, or hot-reload servers that require physical files on disk immediately.…

Google AI Pro – Is standalone Antigravity CLI scripting with external task coordination permitted?

Hi @Lars, Thanks for sharing your architecture before building it. Here is how our current policies and CLI behavior apply to your six questions: No. Per the FAQ and Additional Terms, using agy -p as a secondary worker coordinated by a third-party coding agent like Claude Code isn’t permitted on consumer plans. To run Gemini alongside Claude Code in a multi-agent workflow, use a Gemini API key ("modelProvider": "gemini" in ~/.gemini/antigravity-cli/settings.json with GEMINI_API_KEY) or Gemini…

Cannot republish existing app: "Failed to update Cloud Run service: user has exceeded quota" (quotas healthy, empty domain-mapping request in audit log)

Additional diagnostic information — Los Galácticos FC The application remains accessible at https://losgalacticosfc.ai.studio, but republishing fails with: Failed to update Cloud Run service: user has exceeded quota. Please try again later. The Cloud Run audit logs also show: Domain mapping already exists for losgalacticosfc.ai.studio. Please use a different custom url. Project ID: funky-base-9w2pc Cloud Run service: losgalacticosfc Region: europe-west1 The application builds successfully, and…

Google, Please Address the Removal of Gemini 3.1 Pro (High) from Antigravity!

Hi all, Totally understand the frustration when a model you rely on disappears without warning. Gemini 3.1 Pro was retired on consumer plans and replaced by Gemini 4 Argon (High, Medium, Low) and Gemini 3.8 Flash (High, Medium, Low). For complex coding, debugging, architectural work, and agentic tasks on Google AI Pro, Gemini 4 Argon (High) is the direct upgrade (and you also get access to Claude Sonnet 5.5 and Claude Opus 5.5). The reason you are only seeing Claude 4.6 and GPT-OSS 120B right…

UI Regression: Restore compact layout for Artifacts & Terminal (keep them directly above the input box)

Hi team, The recent workspace layout update has significantly disrupted my developer workflow. Previously, having generated artifacts and terminal execution steps positioned neatly and compactly right above the prompt/input box was extremely ergonomic. Everything was accessible in a single, focused vertical view without visual clutter or unnecessary head/eye movement. In the latest update: Moving terminal execution to a separate bottom console pane Shifting artifacts to an isolated right-side…

Urgent Quota Lockout on New Account

Hello Rebbanpalli, Thanks for your reply. It has been helpful. I have tried age verification; after that, the AI Pro plan is now reflecting both the Antigravity app and IDE. However, I have to say that this is not obvious to know. Antigravity didn’t signal what the issue was (in this case, age verification). Hence, this leads to a lot of frustration and time wasted researching the internet, reading forums, and blindly troubleshooting. In my case, I have an urgent production demo the following…

Cannot republish existing app: "Failed to update Cloud Run service: user has exceeded quota" (quotas healthy, empty domain-mapping request in audit log)

Bug Report — Google AI Studio cannot republish an existing app Hello Google AI Studio Support Team, I am experiencing a persistent publishing issue with my existing application, Los Galácticos FC. Google AI Studio reports that I have exceeded my quota, but my Cloud Run quota checks have not revealed any obvious quota limit being exceeded. I need help identifying and fixing the issue so I can publish my latest changes without losing my existing application or public URL. 1. Project and…

Personal Google AI Ultra: what exactly does "Enable Telemetry = off" stop? (training, retention, human review)

Hi @KEITARO_MARUNO, Here is how each of those works with your current setup: Training: Yes, Enable Telemetry in /config (along with the first-run data-sharing checkbox) is the exact setting referenced in Section 5 of the Antigravity Additional Terms. When it is off, your prompts, code context, tool outputs, and model responses are not collected or used to train models or improve Google products. Retention: With telemetry off, persistent interaction logging is disabled—your prompts and responses…

Hello Google AI Studio Support Team

Hello Google AI Studio Support Team, I am requesting a review of the Gemini API Free Tier eligibility for my Google account and projects. I use the Gemini API for a small internal production management application called ZLINPACK, primarily to extract structured order information from customer documents. The API previously worked, but requests are now failing with HTTP 403 PERMISSION_DENIED and the following message: “The Gemini API free tier is not available for this project. To use the Gemini…

Gemini API Age Requirements for a backend news-classification pipeline

Thank you for the references. Could you help identify the appropriate Google contact or published guidance for this specific backend use case? I found Google’s Syft customer story, which describes using Gemini in Vertex AI to analyze news and deliver personalized news summaries: Google Cloud Syft case study Syft used scalable AI to bring global news to free thinkers, slashing costs and propelling its app to #1 in the process. Syft’s public terms and privacy policy also discuss use by people…

Paid Tier 3 — 403 PERMISSION_DENIED and suspicious API key creation

Title: Paid Tier 3 Gemini API — 403 PERMISSION_DENIED and suspicious API key creation Hello Google Gemini API team, My Gemini API project has stopped working despite having active Paid Tier 3 billing. Project ID: project-b9de73c3-ead4-4c0d-9.. Issues: All Gemini generation requests return 403 PERMISSION_DENIED: Request rejected in accordance with our terms of use and policies. Even a simple “Hello” prompt fails directly in Google AI Studio Playground. Creating a new API key fails with: Failed…

Policy clarification: private monitoring app using official Antigravity CLI and subscription quota

Hi @xorz, How each part of your setup works under our Terms of Service, FAQ, and Headless CLI docs: Section 6 compliance & OAuth handling: Section 6 of the Additional Terms prohibits third-party tools from extracting Antigravity OAuth tokens or wrapping agy inside third-party agents or apps. However, calling the unmodified agy -p --output-format json binary from your own personal local script—where the CLI handles its own login—is supported for direct local scripting. Manual vs. automatic…

Gemini 3.1 Flash TTS: SAFETY / incomplete audio on neutral Spanish narration

Hello Google team, I need help distinguishing a generation/backend failure from a content restriction in gemini-3.1-flash-tts-preview. A short neutral Spanish narration succeeded (10.2 seconds), but an independent 480-word text about maps, scales, colours and terrain returned no audio with finishReason SAFETY and an incomplete-stream message. A longer historical documentary narration had returned the same error previously. We have stopped resubmitting the documentary pending clarification.…

Gemini models are missing from my Antigravity IDE

Hi, I’m using Antigravity IDE on Windows 11 with a Google AI Pro subscription, but I can’t see any Gemini models in the model menu. I only see Claude and GPT-OSS. Antigravity recognizes my plan as Google AI Pro. In Settings → Models, though, I only see the Claude and GPT quota, and there’s no Gemini quota listed. I’ve updated Antigravity, run “Antigravity: Reset Onboarding,” signed out and back in with the Google account linked to my subscription, refreshed the Models page, and restarted the…