Contents
At a glance
- Model ID
- deepseek-v4-flash
- Context length
- 1M tokens
- Max output
- 384K tokens
- Input, cache hit
- $0.007 off-peak to $0.014 peak per 1M tokens (checked August 2026)
- Input, cache miss
- $0.22 off-peak to $0.44 peak per 1M tokens
- Output
- $0.66 off-peak to $1.32 peak per 1M tokens
- Peak hours
- 01:00-04:00 and 06:00-10:00 UTC; off-peak is half the peak rate
- Versus V4 Pro
- Same context and output ceiling at roughly a third of the price
What DeepSeek V4 Flash is
deepseek-v4-flash is the smaller of the two model IDs on DeepSeek's API pricing page. As published in August 2026 it carries the same 1M-token context length and 384K maximum output as deepseek-v4-pro, at roughly a third of the price on every meter.
Same window, same output ceiling, a third of the cost: as with OpenAI's GPT-5.6 tiers, what you give up moving down this lineup is capability rather than room.
The rates
All prices in US dollars, as published August 2026, with the low end of each range being the off-peak rate and the high end peak: cache-hit input $0.007 to $0.014 per million tokens; cache-miss input $0.22 to $0.44 per million; output $0.66 to $1.32 per million.
DeepSeek documents off-peak rates as half the peak rates, with peak hours running 01:00 to 04:00 and 06:00 to 10:00 UTC and everything else off-peak.
What that price does
Off-peak cache-hit input at $0.007 per million is the lowest published input rate discussed anywhere in this glossary. For a workload with a large stable prefix - a fixed system prompt, a fixed tool schema, a fixed reference document - run outside peak hours, the input side of the bill effectively disappears and what you are paying for is output.
That reshapes what is worth doing. Running a second model as a checker, re-running an eval suite nightly on every change, or reprocessing an archive when the prompt improves all become cheap enough not to need a business case. Compare with gemini-2.5-flash-lite at $0.10 input and $0.40 output, which is the closest list-price competitor and has no time-of-day component.
What you should know
The cache-hit rate only applies if the prefix actually matches. Any byte change - a timestamp in the system prompt, a reordered tool list, a session ID - invalidates it and you pay the miss rate, which is 30x higher. If you are choosing this model for its cache economics, audit what varies between your requests before you assume the discount.
And as with any low-cost provider, the non-price questions decide whether it is usable at all for your data. See data residency, zero data retention and DPA. Price is the easy part of that decision and the last part that should be made.
DeepSeek V4 Flash FAQ
How much does deepseek-v4-flash cost?
Cache-miss input is $0.22 to $0.44 per million tokens and output is $0.66 to $1.32, with the low end of each range the off-peak rate. Cache-hit input is $0.007 to $0.014 per million. Published figures as of August 2026.
What is the difference between DeepSeek V4 Flash and V4 Pro?
Price and capability, not size. Both carry a 1M-token context and a 384K output ceiling; Flash is roughly a third of Pro's price on every meter. Test both on your own task rather than assuming the larger model is needed.
Is DeepSeek V4 Flash the cheapest model available?
Its off-peak cache-hit input rate of $0.007 per million is the lowest input figure discussed in this glossary. On cache-miss input at $0.22 to $0.44 it sits alongside gemini-2.5-flash-lite at $0.10 and gpt-5.6-luna at $0.20, so the answer depends heavily on your cache hit rate.
How do I actually get the DeepSeek cache-hit rate?
By keeping the request prefix byte-identical between calls. Any change - a timestamp in the system prompt, a reordered tool list, an injected session ID - invalidates the cache and you pay the miss rate, which is around 30 times higher. Audit what varies before relying on the discount.
Does DeepSeek charge less at night?
It charges less outside two published UTC windows. DeepSeek documents peak hours as 01:00-04:00 and 06:00-10:00 UTC with all other times off-peak at half the rate, so scheduled batch work outside those windows costs half as much regardless of your local time.
Primary source
Checked August 2026. Anything priced or rate-limited moves, so verify against the source before you budget.
Want this built rather than explained?
Knowing what deepseek v4 flash means is one thing; having it running in your business is another. We scope one workflow at a time, quote a fixed price in writing after a free consult, build it, and hand it over with documentation you own.
