---
type: "post"
id: "model-versions-and-lifecycle-gemini-enterprise-agent-platform-6962a5"
slug: "model-versions-and-lifecycle-gemini-enterprise-agent-platform-6962a5"
title: "Model versions and lifecycle: Gemini Enterprise Agent Platform"
description: "Google Cloud’s Gemini Enterprise Agent Platform lifecycle page lists Gemini 4 Hydrogen and Gemini 4 Carbon as released in November 2026, with no retirement dates announced."
retrieval_nugget: "Google Cloud’s Gemini Enterprise Agent Platform lifecycle page lists Gemini 4 Hydrogen and Gemini 4 Carbon as released in November 2026, with no retirement dates announced. For teams choosing a model endpoint, it confirms catalog presence and lifecycle status, but offers no evidence about capabilities, cost, or comparative production performance."
published_at: "2026-11-13"
updated_at: "2026-11-13"
record_date: "2026-11-13"
date_kind: "published_at"
topics: ["gemini","google"]
entities: ["Google Cloud","Gemini Enterprise Agent Platform","Gemini 4 Hydrogen","Gemini 4 Carbon"]
related_patterns: ["harness-architecture-outlives-model-choice"]
novelty_label: "new_fact"
what_changed: "Google Cloud’s model lifecycle page now lists Gemini 4 Hydrogen alongside Gemini 4 Carbon as released."
prior_refs: ["https://newruntime.com/posts/gemini-4-carbon-is-generally-available-on-google-d0a0ba/"]
source_url: "https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/model-versions"
source_urls: ["https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/model-versions"]
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "post:model-versions-and-lifecycle-gemini-enterprise-agent-platform-6962a5"
status: "published"
visuals: []
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
routes: {"html":"https://newruntime.com/posts/model-versions-and-lifecycle-gemini-enterprise-agent-platform-6962a5/","markdown":"https://newruntime.com/posts/model-versions-and-lifecycle-gemini-enterprise-agent-platform-6962a5.md","json":"https://newruntime.com/posts/model-versions-and-lifecycle-gemini-enterprise-agent-platform-6962a5.json"}
markdown_url: "https://newruntime.com/posts/model-versions-and-lifecycle-gemini-enterprise-agent-platform-6962a5.md"
---

# Model versions and lifecycle: Gemini Enterprise Agent Platform

## Retrieval answer

Google Cloud’s Gemini Enterprise Agent Platform lifecycle page lists Gemini 4 Hydrogen and Gemini 4 Carbon as released in November 2026, with no retirement dates announced. For teams choosing a model endpoint, it confirms catalog presence and lifecycle status, but offers no evidence about capabilities, cost, or comparative production performance.

## Summary
Google Cloud’s Gemini Enterprise Agent Platform lifecycle page lists Gemini 4 Hydrogen and Gemini 4 Carbon as released in November 2026, with no retirement dates announced. For teams choosing a model endpoint, it confirms catalog presence and lifecycle status, but offers no evidence about capabilities, cost, or comparative production performance.

## What happened
Google Cloud’s “Model versions and lifecycle” documentation defines lifecycle terms for generally available Gemini and embedding models on Gemini Enterprise Agent Platform. The page separates models promised for at least 12 months after initial release from a group with shorter availability periods. Its stated purpose is to list available models, retirement dates, recommended upgrades, and migration paths.

In the shorter-availability table, Google Cloud lists `gemini-4-hydrogen` with a release date of November 12, 2026. The entry has no announced retirement date and no replacement model ID. The same table lists `gemini-4-carbon`, dated November 10, 2026, likewise with no announced retirement date or replacement. It also lists `gemini-4-carbon-cyber-permissive`, dated November 10, 2026, under the same short-term availability category.

Google Cloud says models in this category remain active until a replacement is released and a retirement date is announced. A replacement launch alone does not retire the existing model: the documentation says the model remains active until Google announces a retirement date. When it does schedule retirement for a short-term model, Google says it will publish a fixed date that gives users at least 45 days to migrate. The page does not assign a maximum lifetime to Hydrogen, Carbon, or the Carbon Cyber Permissive variant.

The lifecycle page also provides a practical migration sequence for models reaching retirement: update an application to point to the recommended upgrade, test all mission-critical features, and deploy the update through the usual process. Google Cloud links separately to its guide for migrating to the latest stable Gemini models. That guidance applies to an eventual replacement workflow, rather than announcing that Hydrogen or Carbon currently requires a migration.

This is a catalog and support-lifecycle update, not a capability launch note. Google Cloud provides model IDs and release dates, but the page does not describe the intended workloads, modalities, context limits, benchmark results, API behavior, pricing, quotas, endpoint locations, or regional availability of Hydrogen or Carbon. It therefore establishes that these model names are listed in the platform’s model catalog; it does not establish which one should be selected for a production workload.

The distinction matters because a release record and a deployment recommendation answer different questions. A team may need a stable named endpoint, a documented replacement path, and enough notice to validate a switch. It may also need latency, quality, safety, tool-use, and cost data. Google Cloud’s table addresses the first set of questions only. The absence of a retirement date is useful operational information, but it is not a stated long-term availability commitment.

## What the archive adds
New Runtime’s earlier coverage of [Gemini 4 Carbon’s general availability](https://newruntime.com/posts/gemini-4-carbon-is-generally-available-on-google-d0a0ba/) on November 11, 2026 described Carbon’s presence on Gemini Enterprise Agent Platform, its Agent Studio configuration, and API access. This lifecycle entry adds a documented November 10 release date for Carbon and places Hydrogen in the same short-term lifecycle category. It is a new fact on a known model-family story, rather than a wholly new platform event.

## Trend signal
This supports, with limits, the idea that [harness architecture outlives model choice](https://newruntime.com/patterns/harness-architecture-outlives-model-choice/). Google Cloud’s table makes model names, replacements, and retirement timing explicit operational variables: applications can outlive a specific endpoint only if their teams can substitute and validate models. The page documents the lifecycle mechanism, not portability of Google Cloud’s harness features, so it neither proves nor resolves the broader question of lock-in.

## Why now
Google Cloud marks the lifecycle page as updated on November 12, 2026, which is also the listed release date for Gemini 4 Hydrogen. Carbon is listed two days earlier, on November 10. The timing suggests that the immediate change is lifecycle visibility in the platform catalog: builders can now see the release records and short-term availability classification together, rather than infer support expectations from a model name alone.

## Why it matters
Model selection is not only a quality comparison. For an agent or application in production, teams need a stable identifier in configuration, an alerting process for lifecycle changes, a tested substitute, and enough validation capacity to determine whether a replacement changes important behavior. Google Cloud’s promise of at least 45 days after it announces a retirement date creates a minimum planning window for short-term models, but it does not guarantee that migration will be easy.

The immediate action is operational rather than evaluative. Record the exact IDs used by each service, watch the lifecycle table, and avoid putting model-specific assumptions throughout application logic. Keep a test suite that covers mission-critical flows so a future endpoint switch can be assessed rather than merely deployed. Hydrogen and Carbon may become viable choices for particular tasks, but this page alone does not provide the capability or price information needed to make that choice responsibly.

## What happens next
- **Google will publish a retirement date for at least one newly listed short-term Gemini 4 model within 12 months.** (12 months, 35%). Watch for: a dated retirement entry in the lifecycle table.

## Limits and uncertainty
Google Cloud’s lifecycle documentation does not provide capability descriptions, benchmarks, pricing, quotas, endpoint locations, regional availability, or migration replacements for Gemini 4 Hydrogen or Gemini 4 Carbon. It also does not explain the practical distinction between Hydrogen and Carbon, or the scope of the Carbon Cyber Permissive variant. “No retirement date announced” describes the page’s current status, not a promise of a fixed support period. The listed release dates come from Google Cloud; this lifecycle page does not independently demonstrate that each model is available to every account, in every location, or for every use case.

## Reader takeaway
Treat this as a lifecycle signal, not a model evaluation. Google Cloud now lists Gemini 4 Hydrogen, Gemini 4 Carbon, and Carbon Cyber Permissive as short-term-availability models without announced retirement dates. If you build on the platform, track the named endpoints and prepare migration tests; assess quality, access, and cost through separate technical documentation and your own workload evaluation.

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- topic hub: [Gemini](https://newruntime.com/topics/gemini.md): Explore the gemini topic hub.
- topic hub: [Google](https://newruntime.com/topics/google.md): Explore the google topic hub.
- collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis.
- collection: [Sources](https://newruntime.com/sources/): See the original sources this archive cites.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)