DeepSeek V4.1 Flash went live on the official API on September 10, 2026 at $0.15 per million input tokens and $0.60 per million output tokens during off-peak hours, double that at peak, with cache hits at $0.003 off-peak. The model name is deepseek-flash, and since October 17, 2026 that name is served by the upgraded DeepSeek-V4.1-Flash-1017 build at the same rates. It replaces V4 Flash outright. DeepSeek also announced that from September 14 it would replace V4 Pro, with every deepseek-v4-pro request served by V4.1 Flash at V4.1 Flash rates, then reversed that on user demand and kept V4 Pro live at its own rates.
That second part is the news for anyone running a DeepSeek workload in production. In a single notice, DeepSeek retired two models and announced that a third would be rerouted onto a new architecture, with four days between the announcement and the switch, before pulling the rerouting back. This page covers the official rate card, how it compares with the models it replaces, what the 29 third-party hosts now serving V4.1 Flash are charging six weeks after launch, and the workload math for buyers deciding whether to stay on the official API or move.
DeepSeek V4.1 Flash official API pricing
Two models are listed on the official API as of October 23, 2026. Both have a 1M-token context window and a 384K maximum output. Prices are USD per million tokens and took effect at 04:00 UTC on September 10. The October 17 upgrade to V4.1-Flash-1017 did not change them.
| Model | Input, cache miss (off-peak / peak) | Input, cache hit (off-peak / peak) | Output (off-peak / peak) | Vision | Concurrency |
|---|---|---|---|---|---|
| deepseek-flash (V4.1-Flash-1017) | $0.15 / $0.30 | $0.003 / $0.006 | $0.60 / $1.20 | Yes | 2,500 |
| deepseek-v4-pro (V4-Pro-0813) | $0.66 / $1.32 | $0.022 / $0.044 | $1.98 / $3.96 | No | 500 |
Peak hours are unchanged: 01:00 to 04:00 and 06:00 to 10:00 UTC, Monday through Friday, with everything else including weekends billed at the off-peak rate. The time-zone breakdown is in DeepSeek peak and off-peak hours.
Three notes on the rate card. The legacy names deepseek-v4-flash and deepseek-v4-flash-vision-exp are still accepted, but the models behind them are retired; requests to either name are served by V4.1 Flash and billed at the Flash price. The V4 Pro row stays live after DeepSeek dropped its plan to route Pro traffic to Flash, covered below. And the concurrency limit on Flash is five times Pro's, which matters for agent fleets more than the per-token price does.
What V4.1 Flash is. DeepSeek describes it as the smallest model in a new architecture family: 552 billion parameters in a mixture-of-experts layout, with a causal encoder-decoder design that activates 8 billion parameters on input and 16 billion on output. Native image understanding is built in rather than bolted on as it was in the V4-Flash-Vision-Exp model. DeepSeek says the KV cache needs a quarter of the HBM and an eighth of the SSD storage of the previous generation, and that its own and third-party tests put V4.1 Flash ahead of V4 Pro on performance, cost, speed, and total task time. The weights are published on Hugging Face with a technical report. On October 17, 2026 DeepSeek swapped the launch build for DeepSeek-V4.1-Flash-1017 behind the same model name, citing stronger benchmark results, faster task completion and shorter outputs, and published the new weights on Hugging Face as well. No per-token price changed.
How the new rate card compares with V4 Flash and V4 Pro
V4.1 Flash is a new model on a new cost curve, not a repricing of an existing one. The table puts its rates next to the two models it replaces, using the off-peak column for all three.
| Token type | V4 Flash (retired) | V4 Pro | V4.1 Flash |
|---|---|---|---|
| Input, cache miss | $0.22 | $0.66 | $0.15 |
| Input, cache hit | $0.007 | $0.022 | $0.003 |
| Output | $0.66 | $1.98 | $0.60 |
The cache-hit rate is the one to look at. DeepSeek's stated reason for the lower cache price is the smaller KV cache, and cache hits are where agent workloads spend most of their input budget. On the official endpoint, OpenRouter's monitoring shows a 93.9 percent cache hit rate for V4.1 Flash in its first hours, which is the kind of traffic that makes the $0.003 rate the number that decides the bill.
The comparison that matters commercially is with V4 Pro, because that is the traffic DeepSeek wanted to move. A V4 Pro buyer paying $0.66 and $1.98 off-peak pays $0.15 and $0.60 for the same requests on V4.1 Flash, a 4.4x gap on input and 3.3x on output, served by a different model that DeepSeek says is better. Since the forced routing was withdrawn, whether it is better for a specific workload is something each buyer can test on their own schedule.
The V4 Pro routing that was announced, then reversed
On September 10, DeepSeek's pricing page said that from 12:00 Beijing time (04:00 UTC) on September 14, 2026, all requests to deepseek-v4-pro would be routed to V4.1 Flash and billed at V4.1 Flash rates until V4.1 Pro was released. The stated reason was that V4.1 Flash "has comprehensively surpassed V4 Pro in performance, cost, speed, and total time."
Before the date arrived, DeepSeek reversed it. The change log now says it will continue providing API services for V4 Pro after September 14 with the billing method unchanged, in response to user demand. V4 Pro stays on the official API at $0.66 input and $1.98 output off-peak, and the V4-Pro-0813 weights stay available from third-party hosts as well.
For buyers the lesson is the notice period, not the outcome. Four days' warning of a forced model swap, then a reversal a few days later, on a model that had been repriced by up to 371 percent three weeks earlier. Anyone running V4 Pro in production lived through three sets of terms in a month without changing a line of code. That is the same structure the August 16 increase produced for V4 Flash, covered in DeepSeek V4 Pro API pricing.
Who hosts DeepSeek V4.1 Flash and at what price
| Host | Input | Output | Cache read |
|---|---|---|---|
| DeepSeek official (off-peak) | $0.15 | $0.60 | $0.003 |
| DeepSeek official (peak) | $0.30 | $1.20 | $0.006 |
| Sail Research | $0.10 | $0.15 | $0.0067 |
| Relace | $0.07 | $0.50 | $0.008 |
| Open Inference | $0.07 | $0.50 | $0.008 |
| inference.net | $0.08 | $0.50 | $0.008 |
| Cloudflare | $0.10 | $0.60 | $0.003 |
| DeepInfra (51% promotion) | $0.098 | $0.294 | $0.0029 |
| Reka AI | $0.15 | $0.60 | $0.003 |
| Together, Fireworks, Baseten, Novita and 8 others | $0.30 | $1.20 | $0.006 |
Six weeks after launch the market has formed. OpenRouter lists 29 providers across 32 standard endpoints for the 1017 build as of October 23, 2026. Twelve of them, including Together, Fireworks, Baseten, Novita, Nebius, Phala and DeepSeek's own listing, sit at exactly DeepSeek's peak rate of $0.30 and $1.20. Below that, a second tier undercuts the official off-peak price: Relace and Open Inference at $0.07 input and $0.50 output, inference.net at $0.08 and $0.50, Cloudflare at $0.10 and $0.60, and Sail Research at $0.10 and $0.15, a quarter of DeepSeek's off-peak output rate. DeepInfra's 51 percent promotion lands at $0.098 and $0.294 while it runs, and GMICloud's 40 percent promotion takes it to $0.18 and $0.72. Those two are promotional prices, so a buyer who depends on them should lock capacity before they reset. Cache reads are where the official API still wins: $0.003 off-peak against $0.006 to $0.008 at most third parties, with only Cloudflare, Reka and Decart matching it.
This is the same path V4 Flash took, only faster. On launch day, September 10, three third parties listed V4.1 Flash, all at 2x to 2.5x the official off-peak rate, anchored on DeepSeek's peak number while they worked out their own costs on a new architecture. V4 Flash had needed months of optimized serving stacks before sixteen of seventeen hosts undercut DeepSeek's off-peak rate. V4.1 Flash got its first hosts below the official off-peak price within six weeks, helped by the October 17 weights release giving hosts a second build to tune for.
The official rate is no longer the floor on list price. It is still the floor on cache reads, and DeepSeek's own endpoint shows 100 percent uptime on OpenRouter against 85 to 100 percent across third parties. The lowest list prices come from smaller hosts with lower throughput, and two of the lowest effective prices are promotions that will end. The table is a snapshot, not a settled market, which is why the Mercatus Token Index tracks the spread rather than any one host.
What a real workload pays
A production assistant doing 10 million input tokens and 1 million output tokens a day, uncached. V4.1 Flash routes are priced at list rates on October 23, 2026; the V4 Flash rows use September 10, 2026 prices. V4 Flash and V4 Pro rows use official off-peak rates for the models being replaced, plus the cheapest third-party host still serving the old V4 Flash weights.
| Route | Daily | Annual |
|---|---|---|
| V4.1 Flash, official, all off-peak | $2.10 | $767 |
| V4.1 Flash, official, uniform traffic (21% peak) | $2.54 | $927 |
| V4.1 Flash, official, all peak | $4.20 | $1,533 |
| V4.1 Flash, hosts at DeepSeek's peak rate | $4.20 | $1,533 |
| V4.1 Flash, Sail Research | $1.15 | $420 |
| V4 Flash, official off-peak (retired Sept 10) | $2.86 | $1,044 |
| V4 Flash 0731 weights, cheapest host | $0.66 | $241 |
| V4 Pro, official off-peak | $8.58 | $3,132 |
Three things fall out of the table.
For V4 Pro traffic, the routing is the whole story. The same 10M/1M workload that costs $8.58 a day on V4 Pro off-peak costs $2.10 on V4.1 Flash off-peak on the official API, with one line changed: the model name. A buyer who validates the new model on their task gets that number. A buyer who cannot, and moves Pro traffic to a third-party host to keep the old model, pays $8.58 or more.
For V4.1 Flash, the official API is no longer automatically the cheapest. A round-the-clock workload pays $2.54 a day on the official API, $4.20 at the twelve hosts anchored on DeepSeek's peak rate, and $1.15 at Sail Research. The third-party discount that took V4 Flash months to form arrived on V4.1 Flash in six weeks, at the low-throughput end of the market.
Cache changes the answer more than the clock does. At a 60 percent cache hit rate, the official off-peak route falls to $1.22 a day, or $445 a year, because 6 million of the 10 million input tokens bill at $0.003 instead of $0.15. The peak and off-peak gap on cached tokens is $0.003, small enough that a cache-heavy agent barely notices the schedule. An uncached workload is where the 2x peak multiplier still bites.
One more row deserves a look. The old V4 Flash 0731 weights are still served by third-party hosts, and the cheapest listed at $0.05 input and $0.16 output on September 10, which runs the workload for $0.66 a day. DeepSeek retired the model; the market did not. For a buyer whose task was fine on V4 Flash and who has no need for the new architecture, the retired model from a third party is a quarter of the price of the new one from DeepSeek. That is what a competitive supply side does when the developer moves on.
Which DeepSeek model and host to use
Running V4 Pro today: test V4.1 Flash on your workload. If it passes, change the model name and the bill drops by about 4x on input and 3x on output. If it fails, stay put; DeepSeek kept V4 Pro on the official API, and third-party hosts serve the weights too.
Running V4 Flash today: you are already on V4.1 Flash. The legacy name routes there and bills at the new rate. Check output quality and the vision behavior if your pipeline sends images, since the encoder is different.
Starting fresh on V4.1 Flash: start on the official API for the cache rate and uptime, then compare the hosts below $0.15 input once your prompt mix is known. An uncached workload that can tolerate lower throughput is cheaper at Sail Research, Relace or Open Inference. A cache-heavy one is still cheapest at DeepSeek.
Cost-sensitive and happy with the old model: V4 Flash 0731 weights from a third-party host, from $0.05 and $0.16. That market will thin out as hosts move capacity to V4.1, so it is not a long-term plan.
Cache-heavy agents: the official $0.003 cache-hit rate is the lowest in the DeepSeek line, matched only by Cloudflare, Reka and Decart among third parties. Structure prompts so the shared prefix hits.
Peak-hour traffic you cannot move: MiniMax M3.1 Flash launched October 19, 2026 at the same $0.15 input and $0.60 output with no peak hours and a 1M context window. A workload that runs inside DeepSeek's peak window pays $4.20 a day on this model and $2.10 on MiniMax.
The model-by-model comparison against GLM, Qwen, and Kimi is in the LLM API pricing comparison.
Where DeepSeek pricing goes from here
DeepSeek's last two pricing events point in opposite directions and make the same case. In August, the official rate went up and the third-party market held the old price, so the official API went from cheapest to among the most expensive within a week. In September, a new model arrived at a lower rate card and the third-party market sat at double the official price on day one, then undercut it within six weeks. In both cases the posted price and the market price were different numbers, and only one of them was set by more than one seller.
What has not changed is how the decisions get made. A seller published a rate, retired two products, and announced the rerouting of a third, with a few days' notice and no instrument for buyers to hold their terms. Better models create competition, and the third-party market on the old weights is proof that buyers now have somewhere else to go. What they still lack is a single price that reflects all of it. How a token exchange works covers how that number gets set by trades instead of by a pricing page, and How to lock in token prices covers what a buyer can do about a routing notice like this one before it lands. The full DeepSeek line, official rates and every host, is kept current in DeepSeek API pricing.
Frequently asked questions
How much does DeepSeek V4.1 Flash cost?
As of October 23, 2026, the official DeepSeek API charges $0.15 per million input tokens and $0.60 per million output tokens off-peak, and $0.30 and $1.20 at peak. Cache hits are $0.003 off-peak and $0.006 at peak. The model name is deepseek-flash. Third-party hosts on OpenRouter range from $0.07 to $0.33 input and $0.15 to $1.32 output.
What happened to DeepSeek V4 Flash?
It was retired on September 10, 2026. The names deepseek-v4-flash and deepseek-v4-flash-vision-exp still work but route to V4.1 Flash and bill at the V4.1 Flash rate. The old V4 Flash weights are still served by third-party hosts.
Is DeepSeek V4 Pro being discontinued?
No. DeepSeek announced on September 10 that deepseek-v4-pro requests would be routed to V4.1 Flash from September 14, then reversed it and kept V4 Pro on the official API at unchanged billing. V4 Pro weights are also available from third-party hosts.
Is V4.1 Flash better than V4 Pro?
DeepSeek says its own and third-party tests put V4.1 Flash ahead of V4 Pro on performance, cost, speed, and total task time, and that is its stated reason for the routing. Buyers with Pro in production can confirm that on their own workloads on their own schedule now that the cutoff is gone.
What is DeepSeek-V4.1-Flash-1017?
The October 17, 2026 upgrade of V4.1 Flash. DeepSeek says it has stronger benchmark results, faster task completion and shorter outputs than the launch build. It serves the same deepseek-flash model name at the same prices, and the weights are on Hugging Face.
Does V4.1 Flash support images?
Yes. Native image understanding is built into the model. The separate V4-Flash-Vision-Exp model is retired.
Are the V4.1 Flash weights open?
Yes. DeepSeek published the weights and a technical report on Hugging Face at release. The upgraded V4.1-Flash-1017 weights followed on October 17, 2026, and 29 providers serve that build on OpenRouter.
Is the official API the cheapest way to run V4.1 Flash?
Not on list price any more. On launch day it was, at every hour, with three third parties at 2x to 2.5x the official off-peak rate. By October 23, 2026 five hosts list below DeepSeek's off-peak rate, led by Sail Research at $0.10 input and $0.15 output, and DeepInfra's 51 percent promotion lands at $0.098 and $0.294. The official API still has the lowest cache-read price and the highest uptime.
Methodology
Official V4.1 Flash and V4 Pro rates, peak-hour definitions, model specifications, the V4 Flash retirement, and the V4 Pro routing rule are from the DeepSeek API documentation pricing page and the September 10, 2026 release notice, read on the day of release. Retired V4 Flash rates are the official off-peak rates in force from August 16 to September 10, 2026. Launch-day third-party prices for V4.1 Flash, and third-party prices for V4 Pro 0813 and V4 Flash 0731 weights, are from OpenRouter's model pages on September 10, 2026. Current V4.1 Flash host prices, cache-read rates, throughput and uptime are from OpenRouter's DeepSeek V4.1 Flash 1017 page on October 23, 2026, standard routing, with promotional prices noted. The V4.1-Flash-1017 upgrade is from DeepSeek's October 17, 2026 release note and API documentation. Workload calculations assume uncached input except where a cache rate is stated; the uniform-traffic case weights peak hours at 35 of 168 weekly hours. Last verified: 2026-10-23.