DeepSeek V4.1 Pro Release Date: There Is No Date, and Three Ways to Know When That Changes

People search for a release date because a vendor usually gives one. DeepSeek has not. The only public sentence about DeepSeek V4.1 Pro is from the 10 September release note that shipped V4.1 Flash: that requests to deepseek-v4-pro would be routed to Flash and billed at Flash rates "until V4.1-Pro launches." That plan was withdrawn after developers objected to a production model being swapped underneath them, and DeepSeek kept serving V4 Pro with billing unchanged. The withdrawal killed the routing, not the name — and no date, window or quarter has been published since.

What exists instead is a shape people keep mistaking for a schedule. In the V4 generation the Pro tier trailed the Flash tier by about two weeks. In V4.1 the Pro name has now trailed Flash by more than seven weeks. That is the only measurable timing fact in the whole story, and it points the opposite way from "imminent."

Why a date is genuinely unknown rather than merely unannounced

An unannounced date still leaves traces: a repository, a changelog branch, a price column, a model card drafted and served behind a flag. None of those exist here.

• No repository. The deepseek-ai organisation's newest entry is DeepSeek-V4.1-Flash, created 10 September 2026. Nothing named for the Pro tier.

• No changelog entry. DeepSeek's API documentation is where every V4 release has been announced; its newest entry is still the September one.

• No price column. The Models & Pricing page lists exactly two models — deepseek-flash and deepseek-v4-pro — with version, context, features, concurrency and rates. A third column is where a launch would first become visible, and there is no third column.

• No benchmark, no card, no parameter count. The 51-page technology report shipped with Flash does not contain the strings "V4.1-Pro" or "V4.1 Pro" anywhere, and its comparison tables use DeepSeek V4 Pro as a baseline the new model beats rather than as a tier it belongs to.

The three checks that end the wait

Any one of these is sufficient, and each is one page load:

• A changelog entry naming DeepSeek V4.1 Pro. This is the earliest reliable signal, and it is the one every previous release has produced.

• A new repository in the deepseek-ai organisation with V4.1-Pro in the name. That is where weights become real.

• A third column on the pricing page. That is where a launch becomes purchasable, and it usually lands after the identifier exists.

Note what is not on the list: a rumour, a screenshot, a "leak" account's countdown, or an inference from the Flash release cadence. This family has one track record — V4 Flash to V4 Pro took about fourteen days — and it has already been contradicted by the current cycle.

What a date would change, and what it would not

A date would change the scheduling conversation and nothing else. DeepSeek V4 Pro is callable today at $0.66 per million input tokens and $1.98 output off-peak, with a 1M context, a 384K output ceiling, MIT weights and a published card. DeepSeek V4.1 Flash is callable at $0.15 and $0.60, with native image input and a cache-hit line at $0.003. Those two cover the flagship and efficiency ends of the line today.

What a V4.1 Pro would add is a third point — presumably a larger backbone on the newer Causal Encoder-Decoder architecture — and its price is the only genuinely unknown variable. DeepSeek's own release note says V4.1 Flash comprehensively surpasses the previous flagship on performance, cost, speed and total runtime, which leaves a Pro tier's case resting entirely on accuracy. That is the one thing this segment publishes last.

What to do while there is no date

Do not schedule a migration around a release note. Both served DeepSeek tiers are on OrcaRouter at DeepSeek's published rates, passed through with no markup on one OpenAI-compatible endpoint — so whichever identifier eventually appears gets added at the vendor's own price on the day it is listed, and the cost of having watched for it is zero.

What to watch, in order: the changelog, the repository, the third column. When one of them moves, the next question is not "is it good" but "what does it cost" — and that answer arrives with the column.