Gemini 2.5 Pro and Flash auto-route on December 1, 2026
On December 1, 2026, Google says API requests to gemini-2.5-pro automatically route to gemini-3.8-flash and are billed at Gemini 3.8 Flash rates. Requests to gemini-2.5-flash automatically route to gemini-3.5-flash-lite and are billed at Gemini 3.5 Flash-Lite rates. The response comes from the replacement model. Both ids are already deprecated, and API access is already limited to people who have used them. gemini-2.5-flash-lite is not deprecated. The deprecations page also says a date in the table is the earliest day a model might be retired.
Last reviewed . Publisher: SunsetDiff.
Steps
Read the two ids. The table is gemini-2.5-pro and gemini-2.5-flash. Each shutdown cell is 2026-12-01. The replacement cell is the id Google names. Open the model page before you change a call.
Pin the replacement and send a test. Call gemini-3.8-flash by that id before December 1, 2026 if your code uses gemini-2.5-pro. Call gemini-3.5-flash-lite by that id before December 1, 2026 if your code uses gemini-2.5-flash. Leaving the old id in the request means Google chooses the replacement and the bill on December 1, 2026.
Read the migration guide before you switch. The 3.8 Flash guide says to replace thinking_budget with thinking_level, and that minimal returns an error on gemini-3.8-flash. It says to remove temperature, top_p, top_k, candidate_count, and prefilled model turns. The 3.5 Flash-Lite checklist says to apply the same parameter rules as 3.6 Flash, and that its default thinking_level is minimal.
Leave gemini-2.5-flash-lite alone. Google says gemini-2.5-flash-lite is not deprecated and remains available through the API until further notice. Its shutdown cell says no shutdown date announced. Do not point those calls at gemini-3.5-flash-lite unless you are choosing a different model on purpose.
What happens on December 1, 2026
Google's caution for gemini-2.5-pro says: "gemini-2.5-pro is deprecated and is no longer available to select in Google AI Studio. Access through the API is limited to existing users who have actively used it in the past. Effective December 1, 2026, all API requests to gemini-2.5-pro will automatically route to gemini-3.8-flash and be billed at Gemini 3.8 Flash rates. Review the breaking API changes in the Gemini 3.8 Flash migration guide before December 1, 2026."
Google's caution for gemini-2.5-flash says: "gemini-2.5-flash is deprecated and is no longer available to select in Google AI Studio. Effective December 1, 2026, all API requests to gemini-2.5-flash will automatically route to gemini-3.5-flash-lite and be billed at Gemini 3.5 Flash-Lite rates. Review the breaking API changes in the Gemini 3.5 Flash-Lite migration guide before December 1, 2026. Access to the 2.5 models is limited to users who have actively used them in the past. gemini-2.5-flash-lite is not deprecated and remains available through the API until further notice."
The table lists shutdown date December 1, 2026 for both ids. The pro replacement cell names gemini-3.8-flash. The flash replacement cell names gemini-3.5-flash-lite. Both cells say the id is no longer available to select in Google AI Studio, and that API requests automatically route on December 1, 2026.
The same page defines shutdown as: "Once a model is shutdown, it is completely turned off, and the endpoint is no longer available." The caution for these two ids is more specific: the request is answered by the replacement model, and the bill uses that model's rates. This guide follows the caution for these two ids.
The page 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 caution says the route is effective December 1, 2026. This guide does not pick which sentence wins if Google sends a later day. Until that notice, the stored shutdown date is 2026-12-01.
The deprecations page says gemini-2.5-pro requests are billed at Gemini 3.8 Flash rates, and gemini-2.5-flash requests are billed at Gemini 3.5 Flash-Lite rates. It links the pricing page for each rate. It does not print a dollar amount in the caution.
The Gemini 3.8 Flash page says: "Gemini 3.8 Flash is available at an introductory rate of $0.75/1M input tokens and $3.75/1M output tokens through December 31, 2026. Standard pricing of $1.50/1M input tokens and $7.50/1M output tokens takes effect January 1, 2027." The Gemini 3.6 page's model table prints $0.30/1M input tokens and $2.50/1M output tokens for gemini-3.5-flash-lite. This guide did not copy cached, batch, or long-context tiers from the pricing page, and it does not compare those figures with the current gemini-2.5-pro or gemini-2.5-flash price.
The pro caution says API access is limited to existing users who have actively used gemini-2.5-pro in the past. The flash caution says access to the 2.5 models is limited to users who have actively used them in the past. A new user can already get HTTP 404. That limit is separate from the December 1 route.
gemini-2.5-flash-lite is a different id. The flash caution says it is not deprecated and remains available through the API until further notice. The stored row has status active, shutdown kind none, and no shutdown_date.
Set the model field to gemini-3.8-flash or gemini-3.5-flash-lite and send one real request before December 1, 2026. That is the test. An old id left in the client is the id Google routes on December 1, 2026, so the test never runs against the model you will be billed for.
The samples below use the Interactions API shape from the Gemini 3.8 Flash quickstart. The second sample keeps that shape and uses the model id from the 3.5 Flash-Lite migration checklist. The 3.6 Flash quickstart uses gemini-3.6-flash, so this guide does not present that sample as a 3.5 Flash-Lite quickstart.
Old id. On December 1, 2026 Google routes this call to gemini-3.8-flash.
from google import genai
client = genai.Client()
interaction = client.interactions.create(
model="gemini-2.5-pro",
input="Write a three.js script that renders a realistic 3D black hole."
)
print(interaction.output_text)
Pinned replacement, from the Gemini 3.8 Flash quickstart
from google import genai
client = genai.Client()
interaction = client.interactions.create(
model="gemini-3.8-flash",
input="Write a three.js script that renders a realistic 3D black hole."
)
print(interaction.output_text)
Same client shape, with the gemini-3.5-flash-lite id from the migration checklist
from google import genai
client = genai.Client()
interaction = client.interactions.create(
model="gemini-3.5-flash-lite",
input="Write a three.js script that renders a realistic 3D black hole."
)
print(interaction.output_text)
The 3.8 Flash checklist says to change the model string to gemini-3.8-flash, and to replace thinking_budget with thinking_level. The values it names are low, medium, and high. Medium is the default. It says: "minimal thinking level is not supported for Gemini 3.8 Flash and will return an error."
The same checklist says to remove temperature, top_p, top_k, and candidate_count. It says candidate_count is unsupported in Gemini 3 and later. It says to remove prefilled model turns and to use previous_interaction_id for multi-turn history. It points at the 3.6 API changes for those parameter rules.
The 3.6 page says: "temperature, top_p, and top_k are deprecated and ignored. In future model generations, supplying these parameters returns an HTTP 400 error." It also says: "API requests ending with a non-empty model role turn are disallowed and return an HTTP 400 Error." The 3.8 page tells you to remove the fields. It does not print a sample response body for a temperature field on gemini-3.8-flash. This guide does not add one.
The 3.5 Flash-Lite checklist says to change the model string to gemini-3.5-flash-lite. It says the default thinking_level is minimal, and to leave minimal for high-volume extraction, routing, or classification. It says to set medium or high for autonomous subagents that use tools, code execution, or multi-step reasoning. It says to apply the same parameter rules as 3.6 Flash.
Both checklists mention function-calling payload shape and thought signatures. The pages to read are linked below. This guide does not restate a rule it did not quote.
Search for gemini-2.5-pro and gemini-2.5-flash. A hit on gemini-2.5-flash-lite is a different model. 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. These two ids are under that count, and the page is published anyway: Gemini API shutdowns on December 1, 2026. Vertex gemini-2.5-pro already has its own retirement date. See why shutdown dates differ by platform.
What happens to gemini-2.5-pro and gemini-2.5-flash on December 1, 2026?
Google says requests to gemini-2.5-pro automatically route to gemini-3.8-flash and are billed at Gemini 3.8 Flash rates. Requests to gemini-2.5-flash automatically route to gemini-3.5-flash-lite and are billed at Gemini 3.5 Flash-Lite rates. The caution says: "gemini-2.5-pro is deprecated and is no longer available to select in Google AI Studio. Access through the API is limited to existing users who have actively used it in the past. Effective December 1, 2026, all API requests to gemini-2.5-pro will automatically route to gemini-3.8-flash and be billed at Gemini 3.8 Flash rates. Review the breaking API changes in the Gemini 3.8 Flash migration guide before December 1, 2026."
Does the request start failing on that date?
The caution says the request automatically routes and the bill changes. It does not say the call starts returning an error on December 1, 2026. The page's general shutdown sentence is: "Once a model is shutdown, it is completely turned off, and the endpoint is no longer available."
Is gemini-2.5-flash-lite deprecated?
No. The flash caution says: "gemini-2.5-flash-lite is not deprecated and remains available through the API until further notice." The table cell says no shutdown date announced.
Can a new API key call these models today?
The pro caution says access is limited to existing users who have actively used gemini-2.5-pro. The flash caution says access to the 2.5 models is limited to users who have actively used them in the past. That limit is already in effect. It is separate from the December 1, 2026 route.
How do I test the replacement before December 1, 2026?
The caution says the route is effective December 1, 2026. The page 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 stored shutdown date stays 2026-12-01 until Google prints a later day.