The Gemini API changelog has an entry headed 8 October 2026 that retires two Flash model IDs by routing them, and that sets a shutdown date for one Deep Research agent. The page does not show a clock time for the entry. The heading is the date Google printed. It is not a claim about the minute the text appeared.
The Flash lines, as Google wrote them
On Gemini 3.7 Flash, the changelog says: “gemini-3.7-flash is deprecated and has been replaced by gemini-3.8-flash. All requests to gemini-3.7-flash are automatically routed to gemini-3.8-flash.” It tells developers to update the model string to gemini-3.8-flash.
On Gemini 3.5 Flash, it says: “gemini-3.5-flash is deprecated and has been replaced by gemini-3.6-flash. All requests to gemini-3.5-flash are automatically routed to gemini-3.6-flash.” The update line there offers gemini-3.6-flash or gemini-3.8-flash.
The model pages repeat the routing. The 3.7 page calls the ID deprecated and says every request is routed to gemini-3.8-flash. The 3.5 page says the same for gemini-3.6-flash, and again offers 3.8 as an alternative string. The 3.8 and 3.6 pages list those IDs as stable.
The deprecations table puts “No shutdown date announced” in the shutdown column for both gemini-3.7-flash and gemini-3.5-flash. The replacement cells are the ones that mention the automatic routing. Google’s note on that page says a shutdown date in the table is the earliest date a model might be retired, and that the exact day is communicated later. For these two IDs there is no such date. The old strings still accept traffic, and that traffic is answered by a different model.
Deep Research Pro preview shuts down on 23 October
The same changelog entry says: “The deep-research-pro-preview-12-2025 agent is deprecated and will be shut down on October 23, 2026.”
It names two replacements. deep-research-preview-04-2026 is “Designed for speed and efficiency, ideal to be streamed back to a client UI.” deep-research-max-preview-04-2026 is “Maximum comprehensiveness for automated context gathering and synthesis.”
The migration line is specific: “To migrate, update the agent parameter in your interactions.create requests from deep-research-pro-preview-12-2025 to deep-research-preview-04-2026 (or deep-research-max-preview-04-2026).”
The deprecations table lists the shutdown as 23 October 2026 and the recommended replacement as deep-research-preview-04-2026 only. The max preview is in the changelog, not in that replacement cell. Both newer agents have “No shutdown date announced” of their own.
Price, on the current pricing page
The pricing page does not list gemini-3.7-flash or gemini-3.5-flash. It does list gemini-3.8-flash and gemini-3.6-flash, and their paid token rates match each other on the standard, batch, flex, and priority tables.
On the standard paid tier, per million tokens in USD, both models show input at $0.75 through 31 December 2026 and $1.50 starting 1 January 2027, and output, including thinking tokens, at $3.75 through 31 December 2026 and $7.50 starting 1 January 2027. Standard input and output are “Free of charge” on the free tier for both. The changelog does not say whether a request that still names the old ID is billed at the destination model’s rate. What it does say is that the request is served by the model it routes to.
So a config that still says gemini-3.7-flash is not a promise that 3.7 answers, and the live price page gives no separate old-versus-new rate to compare. Outputs and latency can change because the model changed. A price change is not established by the current table, because the two destinations publish the same rates and the retired IDs are absent.
What to change
Re-run any eval that pinned gemini-3.7-flash or gemini-3.5-flash. Update the string to the model you actually want: gemini-3.8-flash for the 3.7 route, and gemini-3.6-flash or gemini-3.8-flash for the 3.5 route. There is no announced day when the old IDs stop working.
For Deep Research, 23 October 2026 is a shutdown, not a silent route. Change the agent argument on interactions.create before that date.
This is the Gemini API. It is a different surface from the Gemini app’s plan changes, which we covered when free accounts moved onto Auto. An app subscription row does not tell you which API model ID answered a request.
The Campfire
No commentsNobody has pulled up a log by this one yet. Be the first to say what you make of it.
Held for the desk. It appears after a look.