DeepSeek V4.1 Flash API pricing: peak, off-peak and cache hits

DeepSeek V4.1 Flash is billed per token, and the rate depends on the hour. Peak hours are 01:00–04:00 and 06:00–10:00 UTC, Monday to Friday; every other hour, weekends included, is off-peak at half the peak rates. Input that hits the cache costs a small fraction of fresh input, and there is no charge for writing to the cache. On SeedRouter there is no subscription, and a request that fails is not charged.

The table below is read from the rate table that bills your account, so it shows what a request costs today.

What does DeepSeek V4.1 Flash cost per million tokens?

Off-peak covers 133 of the 168 hours in a week: all of Saturday and Sunday, and every weekday hour outside the two peak windows.

What are DeepSeek's list prices?

For reference, this is the shape of DeepSeek's own prices for the model, from its Models & Pricing page. Each line is a multiple of the peak input rate:

The live table above shows SeedRouter's rate for every line, and the DeepSeek V4.1 Flash page shows them next to DeepSeek's list price. DeepSeek's own table also marks Chinese public holidays as off-peak; SeedRouter applies the weekday schedule every week.

How is a DeepSeek V4.1 Flash request billed?

Every response reports its token counts in usage, and the charge follows them:

• prompt_cache_miss_tokens: prompt tokens read fresh.
• prompt_cache_hit_tokens: prompt tokens served from the cache.
• completion_tokens: the answer plus the reasoning, when thinking is on.

The charge, at the rates of the hour the request runs:

cost = ( cache-miss input × input rate
       + cache-hit input × cache-hit rate
       + completion × output rate ) / 1,000,000

How much does one DeepSeek V4.1 Flash task cost?

Thinking changes the output side most. In our test on the same prime-counting question, thinking off used 2 output tokens, low effort 258 and max 319. Three illustrative shapes of work, in tokens:

Multiply each by the live rates above, and halve it off-peak. Output costs four times input per token, so reasoning and answers usually decide the bill. For an agent that re-reads the same files, cache hits cut the input side to about 2% of fresh input.

Did the DeepSeek V4 Flash price change?

Twice in 2026. On August 16, DeepSeek moved its V4 models to peak and off-peak pricing, with off-peak at half the peak rate. On September 10, when V4.1 Flash replaced V4 Flash, its change log says "API prices have been reduced accordingly". The most recent change was a cut, and the old model names now run on V4.1 Flash at the V4.1 Flash price.

How do I pay less for DeepSeek V4.1 Flash?

• Run flexible work off-peak. Batch jobs, evaluations and backfills cost half outside the peak windows.
• Turn thinking off for simple steps. Classification and extraction rarely need reasoning, and reasoning is output.
• Keep reused context at the start of the prompt. Caching is automatic and needs no setup; repeated prefixes are billed at the cache-hit rate.
• Set max_tokens so a long answer cannot run past what you meant to pay for.

Frequently asked questions

When are DeepSeek V4.1 Flash peak hours?

01:00–04:00 and 06:00–10:00 UTC, Monday to Friday. Everything else is off-peak.

Is there a charge for writing to the cache?

No. DeepSeek V4.1 Flash bills cache misses and cache hits; writing to the cache is not billed separately.

Is there a DeepSeek V4.1 Flash subscription?

Not on SeedRouter. You top up a balance and pay per token, with no plan and no monthly fee.

How do I get started?

The DeepSeek V4.1 Flash API guide shows how to get a key and make the first call.