DeepSeek Pricing 2026: Peak Hours and a Cheaper All-Day Rate

Updated 21 October 2026. DeepSeek's peak hours are 01:00 to 04:00 and 06:00 to 10:00 UTC on weekdays; everything else, including weekends and Chinese public holidays, is half price. SayGM serves V4.1 Flash at $0.10 input and $0.40 output all day, a third below DeepSeek's own off-peak rate of $0.15 and $0.60.

Table of Contents

• Peak and off-peak: how DeepSeek pricing works
• Which DeepSeek model should you use?
• Who serves DeepSeek
• DeepSeek V4.1 Flash pricing by provider
• DeepSeek V4 Flash 0731 pricing by provider
• What 100 million tokens costs
• Running DeepSeek privately
• Switching to SayGM
• Common questions
• TL;DR
• About SayGM

Timing your jobs to off-peak on DeepSeek direct is worth it when the work is batch, the prompt reuses a lot of cached context (cache hits are $0.003 off-peak). For anything that runs during a weekday in Asia or Europe, peak is when your traffic lands, and the all-day rate is the cheaper one.

DeepSeek V4.1 Flash API pricing, checked 21 October 2026

SayGM's rows come from the DeepSeek V4.1 Flash model page and the sealed V4.1 Flash page. DeepSeek's rows come from its pricing page, and everyone else's from OpenRouter's V4.1 Flash listing.

Peak and off-peak: how DeepSeek pricing works

DeepSeek's published rates have two columns. Peak runs 01:00 to 04:00 and 06:00 to 10:00 UTC, Monday to Friday. All other hours are off-peak at half price, including weekends and Chinese public holidays in full.

The two windows are China's working day, so DeepSeek's peak is Beijing's 9 to 6 with a lunch break. Europe and the US move their clocks twice a year and DeepSeek doesn't, so work from UTC there. In US time, Monday's first peak window starts on Sunday evening.

DeepSeek's own pricing table, captured 21 October 2026. Flash costs $0.15 input and $0.60 output off-peak, and $0.30 and $1.20 at peak. Footnote 2 sets the peak hours and adds Chinese public holidays to off-peak.

Peak is 35 hours out of 168, so most of the week is discounted. For a batch job you control the timing of, that is a real 50% saving for no work. For anything user-facing, you pay peak rates during your users' working day and there is nothing to optimise.

The cache-hit rate is the other number worth staring at. At $0.003 per million tokens against $0.15 fresh, DeepSeek discounts repeated context by 98%. No other lab comes close. If your agent resends the same repository or system prompt on every turn, your real input cost is nearer the cache column than the headline.

Which DeepSeek model should you use?

V4.1 Flash, unless the work is hard reasoning or a long coding session, in which case V4 Pro.

There are now two V4.1 Flash checkpoints. The first shipped on 10 September; the second, V4.1-Flash-1017, shipped on 17 October, and it's the one DeepSeek's own API serves. DeepSeek retired the old deepseek-v4-flash name, and requests to it now go to 1017.

Pro costs roughly four times Flash. Flash still scores 91.6% on LiveCodeBench and 86.2% on MMLU-Pro, per the model card.

All four are MIT licensed. That is why two dozen providers serve them and why third-party prices sit below DeepSeek's own.

SayGM carries DeepSeek V4.1 Flash with a sealed route, V4.1 Flash 1017, V4 Flash 0731 with a sealed route, and a sealed V3.2. Not V4 Pro. Each page carries the live rate, throughput and success rate for that model.

Who serves DeepSeek

On 21 October, OpenRouter lists 36 V4.1 Flash endpoints, 23 for the 1017 checkpoint and 27 for the July checkpoint.

SayGM

Good at: the lowest price on V4.1 Flash, sealed routes for the July checkpoint and V3.2, and frontier models on the same key if you want Claude or GPT next to your DeepSeek traffic. Credit is prepaid and drawn down per request, with no subscription and no top-up fee.

Missing: V4 Pro, fine-tuning, and dedicated capacity.

DeepSeek

Good at: the cache discount, which is 98% off a fresh read and the most aggressive in the market. It's also the main route to V4 Pro, and the first to serve each new checkpoint: its API moved to 1017 the day it shipped. It now sells web Search and Fetch APIs at $0.75 per 1,000 successful requests, with billing for calls made through agent tools starting 1 November 2026.

Missing: a flat price. You pay double during peak hours, and requests are processed in China, which some security reviews will ask about.

OpenRouter

Good at: reach. One key gets you every provider in both tables below, sorted by price or speed, with automatic failover when one goes down and filters to exclude providers that train on prompts.

Missing: a price advantage, once the 5.5% card top-up fee is counted, and any privacy of its own. A router reads every prompt that passes through it.

DeepInfra

Good at: being cheap on both checkpoints, with 99.84% uptime on V4.1 Flash.

Missing: a sealed route. They also serve compressed builds of some models, which is worth testing against your own evals.

Fireworks

Good at: what an enterprise checklist asks for. Fine-tuning, dedicated deployments with no rate limits, zero data retention by default, and US-only endpoints.

Missing: a competitive rate. Fireworks charges $0.30 and $1.20 for V4.1 Flash, which is DeepSeek's peak price.

Together

Good at: consolidation, if DeepSeek is one of several open-weight models you run. Dedicated endpoints and reserved capacity, across a catalog past 200 models.

Missing: any DeepSeek-specific advantage. The price mirrors DeepSeek's peak rate.

DeepSeek V4.1 Flash 1017 pricing by provider

1017 is the checkpoint DeepSeek's API now serves, so it's what you get if you call deepseek-flash direct. Third-party pricing on it is a week old and still settling.

SayGM's 1017 route isn't competitive yet. Unless you need 1017 specifically, the September V4.1 Flash at $0.10 and $0.40 is the better buy on SayGM. If you need 1017 sealed, SayGM's sealed 1017 route is half Phala's price.

DeepSeek V4 Flash 0731 pricing by provider

0731 is retired at DeepSeek, which now sends the old model name to 1017, so third-party hosts are the only place left to run it. Prices show it: hosts are clearing capacity. Stay on 0731 only if you've pinned to it and validated against it.

What changed since September

DeepSeek's time-of-day pricing is newer than it looks. Until 16 August 2026, DeepSeek charged one flat rate all day: $0.14 input and $0.28 output for V4 Flash, about $17 per 100M tokens. At 16:00 UTC on 16 August it switched to peak and off-peak billing, launching at $0.22 and $0.66 off-peak and $0.44 and $1.32 at peak. On 23 August (Beijing time) weekends became off-peak all day. Flash has since come down to $0.15 and $0.60 off-peak, but that's still $24 per 100M tokens, and peak is $48. Depending on when your traffic lands, DeepSeek Flash now costs about 40% to 190% more than it did all day in early August. SayGM's $16 all-day rate is close to what DeepSeek used to charge.

This post first went up on 21 September 2026. Five things have changed since.

• DeepSeek shipped V4.1-Flash-1017 on 17 October and switched its API to it. The model name is now deepseek-flash. The old deepseek-v4-flash and deepseek-v4-flash-vision-exp names still work but are served by 1017 at the Flash price.
• Chinese public holidays are now listed as off-peak. Peak is still 01:00 to 04:00 and 06:00 to 10:00 UTC on weekdays, and the rates are unchanged at $0.15 and $0.60 off-peak, $0.30 and $1.20 at peak.
• SayGM's V4.1 Flash rate rose from $0.089 and $0.356 to $0.10 and $0.40, or from $14 to $16 per 100M tokens. It's still a third below DeepSeek's off-peak rate.
• SayGM added a sealed V4.1 Flash route at $0.27 and $1.08, $43 per 100M. That's now the cheapest sealed DeepSeek on SayGM. Sealed 0731 went from $0.343 and $1.028 to $0.418 and $1.254, and sealed V3.2 went from $0.431 to $0.95 flat.
• The third-party market moved. A few smaller providers now undercut the major hosts on V4.1 Flash, Morph's 55% promotion has ended, and Fireworks rose from $0.22 and $0.66 to DeepSeek's peak price.

DeepSeek also launched standalone Search and Fetch APIs at $0.75 per 1,000 successful requests. Calls made through DeepSeek Harness, Claude Code or Codex are free until 1 November 2026, then billed.

Running DeepSeek privately

SayGM serves four sealed DeepSeek routes inside Intel TDX enclaves:

• V4.1 Flash at $0.27 and $1.08
• V4.1 Flash 1017 at $0.27 and $1.08
• V4 Flash 0731 at $0.418 and $1.254
• V3.2 at $0.95 flat

The prompt is sealed for the whole request, so nobody reads it, including us. Each session returns an attestation report, a signed statement from the chip confirming what ran; our guide to private inference covers how to check one.

At $43 per 100M tokens, sealed V4.1 Flash is the cheapest of the four. Phala is the other enclave provider, and its promotional V4.1 Flash rate of $0.21 and $0.84 ($34) is cheaper than SayGM's sealed route while the promotion lasts. On 1017, Phala is $0.75 and $1.60, so SayGM is half the price there.

DeepSeek is a Chinese company and requests to its own API are processed there. On a sealed route that question doesn't arise, because the prompt is unreadable to every party in the chain.

Switching to SayGM

A base URL and a model ID:

from openai import OpenAI

client = OpenAI(
    base_url="https://api.saygm.com/v1",
    api_key="YOUR_SAYGM_KEY",
)

response = client.chat.completions.create(
    model="deepseek-v4.1-flash",
    messages=[{"role": "user", "content": "Summarise this changelog for a release note."}],
)
print(response.choices[0].message.content)

Swap in deepseek-v4.1-flash-tee for the sealed route, or deepseek-v4.1-flash-1017 to match DeepSeek's current checkpoint. The quickstart covers keys, and every SayGM rate in this post comes from the model pages, checked 21 October 2026, which carry the current rate if prices move.

Common questions

When are DeepSeek's peak hours?

DeepSeek's peak hours are 01:00 to 04:00 and 06:00 to 10:00 UTC, Monday to Friday, as of October 2026. All other hours are off-peak at half price, including weekends and Chinese public holidays.

What are DeepSeek's peak and off-peak hours in IST?

In India, DeepSeek's peak hours are 06:30 to 09:30 and 11:30 to 15:30 IST, Monday to Friday, as of October 2026. Every other hour is off-peak, including all of Saturday and Sunday.

How much does the DeepSeek API cost per million tokens in 2026?

As of 21 October 2026, DeepSeek charges $0.15 per million input tokens and $0.60 output for Flash off-peak, doubling to $0.30 and $1.20 at peak. Cache hits are $0.003 off-peak. Third-party hosts go lower: V4.1 Flash is $0.10 and $0.40 on SayGM, which is $16 per 100 million tokens against $24 off-peak and $48 at peak from DeepSeek direct.

How much does DeepSeek V4 Pro cost?

As of October 2026, DeepSeek V4 Pro costs $0.66 per million input tokens and $1.98 output off-peak, and $1.32 and $3.96 at peak, on DeepSeek's own API. It's about four times the price of Flash. SayGM doesn't serve V4 Pro.

How much does it cost to fine-tune DeepSeek?

DeepSeek's own API doesn't offer fine-tuning as of October 2026; you fine-tune through a host. Fireworks' serverless training API charges $4.98 per million training tokens for V4.1 Flash, and its managed LoRA training is $10 per million training tokens for models over 300B parameters, which includes V4.1 Flash.

What is the difference between V4 Pro and V4 Flash?

Pro is 1.6 trillion parameters with 49 billion active, built for hard reasoning and long coding work. Flash is 284 billion with 13 billion active in the July checkpoint, or 552 billion with 8 to 16 active in V4.1. Pro costs about four times more. Most production traffic runs fine on Flash.

Is DeepSeek cheaper than Kimi or GLM?

Mostly, as of October 2026: V4.1 Flash at $16 per 100 million tokens on SayGM is cheaper than every Kimi model and every full-size GLM. The exception is GLM 5.3 Flash, at $0.07 and $0.23, or about $10. Our Kimi pricing guide, GLM pricing guide and cheapest LLM API comparison have the full numbers.

Is there a free DeepSeek API?

Not as of 21 October 2026, either from DeepSeek or from any provider on OpenRouter's V4.1 Flash listing. At $0.10 per million input tokens on SayGM, a million tokens of V4.1 Flash costs ten cents, which is close enough to free for testing.

TL;DR

DeepSeek's peak hours are 01:00 to 04:00 and 06:00 to 10:00 UTC on weekdays, and all other hours are half price. Its API now serves the 17 October checkpoint at $0.15 and $0.60 off-peak. SayGM serves V4.1 Flash at $0.10 and $0.40 all day, a third below that, and adds a sealed route at $0.27 and $1.08.

Point your OpenAI client at api.saygm.com/v1, set the model to deepseek-v4.1-flash, and say gm to AI.

About SayGM

SayGM is a drop-in inference gateway for teams who don't want to just take a company's word that their prompts are private. Route to a confidential model and your prompt is sealed inside hardware-verified execution, where not even SayGM can see it. That's not a policy, it's provable. Swap in your existing OpenAI, Anthropic, or Gemini code and you're covered in minutes, at transparent, published rates with no hidden markup.

Say gm to AI at saygm.com.

Website | Twitter | Discord | Blog | Medium | Docs