Google Removes the Admin Toggle for Gemini 3.5 Flash — Here's What IT Teams Need to Know
As of June 9, 2026, Gemini 3.5 Flash is the permanent, non-disableable default in Gemini Enterprise. The admin toggle is gone for good.
Google made a quiet but consequential change to Gemini Enterprise over June 8 and 9, 2026. Gemini 3.5 Flash is now the permanent default model for the Gemini Enterprise app, and the Feature Management toggle that allowed admins to disable it has been removed. Permanently.
If your organisation runs Gemini Enterprise in the Global, US, or EU multi-regions, this change is already in effect and there is no way to roll it back.
What exactly changed
Until this week, Google Cloud admins could navigate to the Gemini Enterprise app settings in the Google Cloud console, find the Feature Management tab, and toggle Gemini 3.5 Flash off for their users. That option is gone.
From June 9, 2026:
- Gemini 3.5 Flash is enabled by default for every user in the Gemini Enterprise app
- The Feature Management toggle for this model no longer appears in the console
- Admins cannot disable it
The change rolled out across the Global, US, and EU regions. It was originally scheduled for June 8, then pushed by one day to June 9 before taking effect.
What happened to Gemini 2.5 Flash
It has been removed from the model selector. Because Gemini 3.5 Flash is now the current GA Flash model, Gemini 2.5 Flash no longer appears as a selectable option in either the Gemini Enterprise app assistant or Agent Designer.
Users can still access Gemini 2.5 Pro and Gemini 3.5 Flash via the model selector dropdown, but if anyone in your organisation was relying on 2.5 Flash specifically, that path is closed in the UI.
For the Business edition, the same applies: Gemini 3.5 Flash has replaced the Gemini 2.5 Flash option in the model selector.
What happened to Agent Designer agents
Google handled this automatically. Agents in Agent Designer running in the US and Global regions that had previously been migrated from Gemini 2.5 Flash or Gemini 2.5 Pro to Gemini 3.1 Pro have now been migrated again, this time to Gemini 3.5 Flash. No action was required from admins or developers for that migration to happen.
If you want those agents to use a different model, you can change it by editing the agent’s settings in either conversational chat or the flow builder.
New agents created in Agent Designer default to Gemini 3.1 Pro in the US and Global regions. After creation, you can switch them to Gemini 2.5 Pro or Gemini 3.5 Flash in the agent settings.
Why this matters beyond the toggle
The removal of the toggle is the most visible part of this change, but the broader picture is worth understanding.
Google’s model retirement pace is accelerating. The gemini-2.0-flash-001 and gemini-2.0-flash-lite-001 API endpoints shut down on June 1, 2026. Gemini 2.5 Flash, the model that just disappeared from the selector, carries an October 16, 2026 shutdown date. That means any team now pointing API calls at gemini-2.5-flash is already on the clock for a second mandatory migration, roughly four months away.
Production systems with hardcoded model identifiers are carrying real risk. A model string that works today may return errors before the end of the year. This is not hypothetical: the June 1 shutdowns already caught some teams out.
Google’s position is that Gemini 3.5 Flash delivers near-Pro intelligence at Flash-tier cost and speed, including Pro-level coding performance and parallel agentic execution. The forced migration is Google pushing organisations toward what it considers the current capable baseline, not a cost-cutting measure on Google’s side.
What you should do now
Check your API integrations first. If you have services calling Gemini models by specific versioned identifiers, audit them. Treat hardcoded model IDs as technical debt that needs active management. The pace of model retirements means this is no longer a set-and-forget configuration.
Review your Agent Designer agents. The automatic migration to Gemini 3.5 Flash has already happened for affected agents. Verify that your agents are behaving as expected and, if a specific model was chosen for a reason (output style, latency, cost), revisit whether Gemini 3.5 Flash meets those needs or whether switching to Gemini 2.5 Pro is more appropriate.
Plan for the October cutover. Gemini 2.5 Flash has a confirmed shutdown date of October 16, 2026. If your team was considering whether to migrate to the 3.x model tier, that decision has been made for the UI layer already. For anything touching the API, build that migration into your roadmap now rather than responding to a deadline.
Know where your remaining controls live. The Feature Management toggle for Gemini 3.5 Flash is gone, but other admin controls remain. Centralised Gemini Enterprise settings, including service on/off controls and data sharing configurations, are still available in the Generative AI section of the Workspace Admin console. Granular control over Workspace data access is still governed by your licence.
EU organisations: Gemini 3.5 Flash maintains compliance with data residency and machine learning processing requirements in the EU. The model change does not affect your data residency posture.
The bottom line
This is Google consolidating around its current model generation and removing the ability for organisations to stay on older defaults indefinitely. The technical migration for most users was automatic. The operational question is whether your API integrations, custom agents, and internal tooling have kept pace, and whether your team is tracking the October 2025 deadline for the next forced cutover.
The Gemini Enterprise release notes and Agent Platform release notes are worth bookmarking if you are not already watching them. Model changes at this cadence mean release notes have become a routine part of enterprise AI operations, not just developer reading.
Updates to this story
31 July 2026: Gemini 3.5 Flash removed from global region in Gemini Enterprise app on August 4
The story has moved into a second stage. The original post covered Google removing the admin toggle for Gemini 3.5 Flash; this update concerns the model itself disappearing from the in-app model selector.
According to Google Cloud release notes published July 27, Gemini 3.5 Flash will be removed as a selectable model from the global region in the Gemini Enterprise web app on August 4, 2026. Tenants whose workloads are pinned to that model in the global region need to migrate before that date. This is not a full model retirement — Google’s lifecycle page puts the 3.5 Flash retirement horizon at least 12 months from its May 19 release — but it is a hard regional availability cutoff.
The designated replacement is Gemini 3.6 Flash, which reached general availability on July 21, 2026. The migration is not a simple swap. Moving to 3.6 Flash requires removing temperature, top_p, top_k, and candidate_count from API calls, replacing thinking_budget with thinking_level, and retesting function calling and multi-turn state management.
Fine-tuning workloads face additional constraints: supervised and reinforcement fine-tuning for 3.5 Flash is now restricted to us-central1 and europe-west4, with tuned model serving limited to US and EU multi-region endpoints.
The deadline is tight. If your tenant targets the global region, treat August 4 as a hard cutoff.
16 June 2026: Gemini Enterprise Gemini 3.5 Flash admin toggle removed on June 16 — corrected date confirmed
Google has formally corrected the removal date for the Gemini 3.5 Flash admin toggle. According to the Gemini Enterprise release notes, the toggle was definitively removed on June 16, 2026, not June 8 or June 9 as stated in earlier notices. The original article’s date was drawn directly from Google’s own June 5 release note, which Google has since flagged as incorrect.
The current state is straightforward: the feature management toggle no longer exists. Gemini 3.5 Flash is permanently enabled by default for all users in the Gemini Enterprise app and cannot be disabled. This applies across the Global, US, and EU multi-regions.
The functional outcome described in the original article remains accurate. IT teams have no administrative path to revert the model or suppress it at the tenant level. What changes with this update is the confirmed hard date. June 16, 2026 is now the official, corrected date on record from Google.
If your organization documented the June 8 or June 9 date in internal change logs or compliance records, those should be updated to reflect June 16, 2026 as the authoritative removal date per the official release notes.