Google Gemini 3.7 Flash and 2.5 Flash Image shutdowns from January 28, 2027 to March 15, 2027
The dataset has 3 rows in this window. Google Vertex AI gemini-3.7-flash on 2027-01-28. Google Gemini API gemini-2.5-flash-image on 2027-03-15. Google Vertex AI gemini-2.5-flash-image on 2027-03-15. Each row stores status deprecated and an official replacement. Vertex gemini-3.7-flash stores gemini-3.8-flash. Both gemini-2.5-flash-image rows store gemini-3.1-flash-lite-image on the same provider. A deprecation date is Not published on these rows. The Vertex model versions page says retirement dates will not move earlier. The Gemini API deprecations page says a table date is the earliest possible retirement, and that a shutdown turns the endpoint off. Gemini API gemini-3.7-flash is a different row: it stores no shutdown_date.
Last reviewed . Publisher: SunsetDiff.
Steps
Read the three rows. The table is these 3 ids. Shutdown date and replacement are the fields stored on each row. Not published means the field is empty. Open the model page before you change a call. Vertex and the Gemini API each have their own gemini-2.5-flash-image row.
Swap the model id. On Vertex, change gemini-3.7-flash to gemini-3.8-flash. The 3.8 Flash guide says: "To migrate to 3.8 Flash from 3.7 Flash or 3.6 Flash, do the following steps." Then: "Change model string in your API calls to gemini-3.8-flash." On each provider, change gemini-2.5-flash-image to gemini-3.1-flash-lite-image. Do not send the Vertex id to the Gemini API host, or the Gemini API id to Vertex.
Test before the date. Send one real request to gemini-3.8-flash before January 28, 2027. Send one real image request to gemini-3.1-flash-lite-image on the provider you use before March 15, 2027. The Vertex page says to test mission critical features. The [pre-migration test checklist](/guides/pre-migration-test-checklist/) is the longer list.
Read the stored region note. These rows do not store a region list. The stored note is a release date. A location list for gemini-3.7-flash, gemini-2.5-flash-image, or gemini-3.1-flash-lite-image is Not published on these rows. The 3.8 Flash quickstart sets location to global. That sample is not a region list for the retiring ids.
Models and dates
These 3 rows are the Google ids in this guide. The model id links to its SunsetDiff model page. Provider, deprecation date, shutdown date, and replacement are the fields stored on that row. Not published means the field is empty. These rows do not store a deprecation_date.
Vertex puts gemini-3.7-flash on the shorter-availability table. The stored excerpt is "gemini-3.7-flash | August 13, 2026 | January 28, 2027 | gemini-3.8-flash". The page says: "Short-term availability models remain active until a replacement model is released and a retirement date is announced. When we schedule a model for retirement, we post a fixed date in the following table that gives you at least 45 days to migrate." And: "Even after a replacement launches, a model remains active until we announce its retirement date." The stored release note is: "Release date on the Vertex model versions page: 2026-08-13.".
Vertex puts gemini-2.5-flash-image on the image-model table under models available for at least 12 months after release. The stored excerpt is "gemini-2.5-flash-image | October 2, 2025 | March 15, 2027 | gemini-3.1-flash-lite-image". The stored release note is: "Release date on the Vertex model versions page: 2025-10-02.".
The Gemini API lists gemini-2.5-flash-image under Nano Banana models. The stored deprecations excerpt is "gemini-2.5-flash-image | October 2, 2025 | March 15, 2027 | gemini-3.1-flash-lite-image". The pricing excerpt says: "Warning: Gemini 2.5 Flash Image (gemini-2.5-flash-image) is deprecated and will be shut down on March 15, 2027; migrate to Gemini 3.1 Flash Lite Image to avoid service disruption." The stored release note is: "Release date on the deprecations page: 2025-10-02.".
Gemini API gemini-3.7-flash is not one of the three shutdown rows. Status deprecated. Shutdown kind none. No shutdown_date is stored. The stored note says: "Release date on the deprecations page: 2026-08-13. Deprecated, requests auto-routed to gemini-3.8-flash, no shutdown date.". The deprecations excerpt says the endpoint is not shutting down and automatically routes API requests to gemini-3.8-flash. Do not copy the Vertex date 2027-01-28 onto that API id.
3 model ids with an exact shutdown date on 2027-01-28 or 2027-03-15.
On January 28, 2027 the Vertex gemini-3.7-flash row reaches the shutdown date stored in the dataset. On March 15, 2027 both gemini-2.5-flash-image rows reach the shutdown date stored on that row. The Vertex model versions page says: "if you don't have time to follow the guide and just need to quickly resolve the errors caused by models reaching their retirement dates, do the following:" The three steps on that page are: update the application to point to the recommended upgrades, test all mission critical features, and deploy the updates. It does not print an HTTP status for that moment.
The same page says: "While retirement timelines may be extended, they won't be moved to an earlier date than what is listed." And: "Listed retirement dates refer to retirement of support in Gemini Enterprise Agent Platform. Models may remain accessible through the Gemini API after these dates have passed." A Vertex retirement date is not a Gemini API shutdown date. The Gemini API deprecations page says: "Once a model is shutdown, it is completely turned off, and the endpoint is no longer available." It also says: "The shutdown dates listed in the table indicate the earliest possible dates on which a model might be retired. We will communicate the exact shutdown date to users with advance notice to ensure a smooth transition to a replacement model." The Gemini API gemini-2.5-flash-image row stores shutdown_date_kind exact. The page still calls the table date the earliest possible retirement.
The pricing warning for the Gemini API image id says: "Warning: Gemini 2.5 Flash Image (gemini-2.5-flash-image) is deprecated and will be shut down on March 15, 2027; migrate to Gemini 3.1 Flash Lite Image to avoid service disruption." It says "service disruption" and does not print an HTTP status.
This section is the Vertex id. The 3.8 Flash guide says: "To migrate to 3.8 Flash from 3.7 Flash or 3.6 Flash, do the following steps." Then: "Change model string in your API calls to gemini-3.8-flash." Then: "Replace integer thinking_budget configurations with the string enum thinking_level". thinking_level="MINIMAL" is not available for 3.8 Flash. Explicitly setting thinking_level to MINIMAL will return an API validation error. Valid configurations are LOW, MEDIUM (default), and HIGH. Under mandatory API rules, that page says deprecated sampling parameters (temperature, top_k, top_p): "These parameters are ignored by the backend." It says unsupported parameters: "Passing frequency_penalty, presence_penalty or candidate_count will throw an active API error." The checklist says: "Remove the following unsupported parameters: frequency_penalty, presence_penalty, candidate_count, temperature, top_p, and top_k." It also says: "History payloads cannot end with a "model" role turn." And: "Prefilled model turns are unsupported and must be removed when upgrading."
The stored replacement row is gemini-3.8-flash, status active, shutdown kind not_before, shutdown date 2027-09-02. The model versions excerpt says "gemini-3.8-flash | September 2, 2026 | September 2, 2027 or later". September 2, 2027 or later is a floor for the replacement. It is not the 2027-01-28 date.
The quickstart below sets location to global. A location list for gemini-3.7-flash is Not published on the stored row. Price, context window, and max output tokens are Not published on the Vertex gemini-3.7-flash row and on the Vertex gemini-3.8-flash row. The migrate diff shows the stored model-id swap.
Retiring Vertex model id, same client shape as the 3.8 Flash quickstart
from google import genai
client = genai.Client(enterprise=True, project="PROJECT_ID", location="global")
response = client.models.generate_content(
model="gemini-3.7-flash",
contents="How does AI work?",
)
print(response.text)
3.8 Flash quickstart
from google import genai
client = genai.Client(enterprise=True, project="PROJECT_ID", location="global")
response = client.models.generate_content(
model="gemini-3.8-flash",
contents="How does AI work?",
)
print(response.text)
Change gemini-2.5-flash-image to gemini-3.1-flash-lite-image
Each provider stores its own replacement id, gemini-3.1-flash-lite-image. The display name on the Vertex docs excerpt is Gemini 3.1 Flash-Lite Image (Nano Banana 2 Lite). The Gemini API pricing page heading is Gemini 3.1 Flash Lite Image (Nano Banana 2 Lite).
The Vertex replacement stores status active, shutdown kind not_before, and shutdown date 2027-06-28. The model versions excerpt says "gemini-3.1-flash-lite-image | June 23, 2026 | June 28, 2027 or later". June 28, 2027 or later is a floor. It is not a shutdown on March 15, 2027.
The Gemini API replacement stores status active, shutdown kind none, and no shutdown_date. The deprecations excerpt says "gemini-3.1-flash-lite-image | June 30, 2026 | No shutdown date announced".
The Vertex image migration record says: "Both ids are image models. The Gemini API image generation guide's text-to-image sample uses client.interactions.create and POST https://generativelanguage.googleapis.com/v1beta/interactions. These records are Vertex model ids, and that guide does not show a Vertex host, so this page does not show a model-id swap." This guide does not invent a Vertex image call. The Gemini API migration diff stores a text-to-image sample that calls client.interactions.create. The model field is the only change shown below. The key header on that sample is left out. The Vertex migrate diff and the Gemini API migrate diff are the stored swaps.
Gemini API text-to-image sample shape, retiring model id
from google import genai
client = genai.Client()
interaction = client.interactions.create(
model="gemini-2.5-flash-image",
input="Create a picture of a nano banana dish in a fancy restaurant with a Gemini theme",
)
Same sample shape, stored Gemini API replacement
from google import genai
client = genai.Client()
interaction = client.interactions.create(
model="gemini-3.1-flash-lite-image",
input="Create a picture of a nano banana dish in a fancy restaurant with a Gemini theme",
)
Context tokens and max output tokens below are the numbers stored on each row. Not published means that field is empty. Input price and output price are the stored input_price_per_mtok and output_price_per_mtok fields. A blank field is not a price of zero.
The Vertex pricing excerpt for gemini-2.5-flash-image says: "Gemini 2.5 Flash Image Input (text, image)*** $0.3 per 1M tokens for the <= 200K column" The excerpt for gemini-3.1-flash-lite-image says: "Gemini 3.1 Flash-Lite Image (Nano Banana 2 Lite) Input (text, image, video) [3] $0.25 per 1M tokens for the <= 200K column" Those excerpts quote the input cell. An output price sentence is Not published in the stored excerpts. The dataset still stores output_price_per_mtok 2.5 on Vertex gemini-2.5-flash-image and 1.5 on Vertex gemini-3.1-flash-lite-image.
The Gemini API pricing page's paid standard input cell for gemini-2.5-flash-image says $0.30 (text / image). The paid output cell says $0.039 per image. The dataset stores output_price_per_mtok 0.039 on that row. This guide does not relabel $0.039 per image as a per-million-token price. The replacement's paid input cell says $0.25 (text/image/video). Its paid output cell says $1.50 (text and thinking) and $30.00 (images). The same cell says: "Equivalent to $0.0336 per 1K resolution image". The dataset stores output_price_per_mtok 1.5 on that replacement, which matches the text-and-thinking figure, not the $30.00 (images) figure.
Vertex gemini-3.7-flash and Vertex gemini-3.8-flash store no price. Those cells say Not published.
Stored context, max output, and price fields. Not published means the field is empty.
A region or location list is Not published on these rows. The stored notes are release dates. Vertex gemini-3.7-flash: "Release date on the Vertex model versions page: 2026-08-13.". Vertex gemini-2.5-flash-image: "Release date on the Vertex model versions page: 2025-10-02.". Gemini API gemini-2.5-flash-image: "Release date on the deprecations page: 2025-10-02.".
The Vertex page says: "Listed retirement dates refer to retirement of support in Gemini Enterprise Agent Platform. Models may remain accessible through the Gemini API after these dates have passed." The 3.8 Flash quickstart in this guide sets location to global. That is the sample's location argument. It is not a published region list for gemini-3.7-flash. November and December 2026 Vertex locations for other ids are on Vertex AI Gemini retirements in November and December 2026. This guide does not copy those lists onto these ids.
Search the repository for the three ids in the first table. A hit is a candidate. Also search configs that set a Vertex location, and configs that call the Gemini API host. The same model id string can point at either provider. How to find every model ID in your codebase covers SDK argument names. The scan page builds a search from dataset ids that have a shutdown date.
SunsetDiff publishes a shutdown batch page when at least 3 model records share one provider and one exact shutdown date, or when a date is explicitly included.
Which Google models shut down in January and March 2027?
The dataset lists Google Vertex AI gemini-3.7-flash on 2027-01-28; Google Gemini API gemini-2.5-flash-image on 2027-03-15; Google Vertex AI gemini-2.5-flash-image on 2027-03-15. Each row is status deprecated. A deprecation date is Not published.
What is the official replacement?
The recommended replacement is the official replacement stored on each row. gemini-3.7-flash: gemini-3.8-flash. gemini-2.5-flash-image: gemini-3.1-flash-lite-image. gemini-2.5-flash-image: gemini-3.1-flash-lite-image. Vertex gemini-3.8-flash stores shutdown date 2027-09-02 with kind not_before. Vertex gemini-3.1-flash-lite-image stores 2027-06-28 with kind not_before. Gemini API gemini-3.1-flash-lite-image stores no shutdown_date and the excerpt says "No shutdown date announced".
What breaks on the shutdown date?
The Vertex page says to resolve the errors caused by models reaching their retirement dates. It does not print an HTTP status. It says: "While retirement timelines may be extended, they won't be moved to an earlier date than what is listed." The Gemini API page says: "Once a model is shutdown, it is completely turned off, and the endpoint is no longer available." The pricing warning says: "Warning: Gemini 2.5 Flash Image (gemini-2.5-flash-image) is deprecated and will be shut down on March 15, 2027; migrate to Gemini 3.1 Flash Lite Image to avoid service disruption."
Does Gemini API gemini-3.7-flash shut down on January 28, 2027?
No. The Vertex row stores 2027-01-28. The Gemini API row stores shutdown kind none and no shutdown_date. Its excerpt says the endpoint is not shutting down and automatically routes API requests to gemini-3.8-flash.
Do Vertex and the Gemini API share one March 15 page?
No. A region or location list is Not published on these rows. The stored note on each row is a release date. The 3.8 Flash quickstart sets location to global. That sample is not a region list for the retiring ids.
What changes besides the gemini-3.7-flash model id?
The 3.8 Flash guide says: "Replace integer thinking_budget configurations with the string enum thinking_level". thinking_level="MINIMAL" is not available for 3.8 Flash. Explicitly setting thinking_level to MINIMAL will return an API validation error. Valid configurations are LOW, MEDIUM (default), and HIGH. It says: "These parameters are ignored by the backend." It also says: "Passing frequency_penalty, presence_penalty or candidate_count will throw an active API error." The checklist says: "Remove the following unsupported parameters: frequency_penalty, presence_penalty, candidate_count, temperature, top_p, and top_k." And: "Prefilled model turns are unsupported and must be removed when upgrading." Price and token limits for the two Vertex text rows are Not published.
What token limits change for the image ids?
Vertex gemini-2.5-flash-image stores context 32768 and max output 32768. Vertex gemini-3.1-flash-lite-image stores context 65536 and max output 4096. Gemini API gemini-2.5-flash-image stores context 65536 and max output 32768. Gemini API gemini-3.1-flash-lite-image stores context 65536 and max output 4096.