Cheapest way to run DeepSeek V4.1 Flash: providers compared
Compare DeepSeek V4.1 Flash API prices across every gateway and plan, with live widgets for the cheapest endpoints and the best monthly plan.
DeepSeek shipped V4.1 Flash on 2026-09-10 and moved its flagship traffic onto it: from 2026-09-14, every V4 Pro request is served by V4.1 Flash at Flash rates. The model takes text and images, holds a one-million-token context, and bills on a peak and an off-peak rate. The same weights cost different amounts on each gateway, so the endpoint you pick sets the bill.
The cheapest endpoints
Providers list this model in one of two shapes: a single flat rate, or a pair of peak and off-peak rates. The three tables below rank each shape on its own, one row per plan. They start on Effective prices; turn the toggle off for the raw listed rate. All three read the same data as the main table.
Flat-rate endpoints
Providers that publish one rate, with no time window.
| Plan | Pricing /1M | Privacy | Performance | Monthly quota | |||
|---|---|---|---|---|---|---|---|
| Input | Output | Cached read | Logs | Trains | |||
| Portal PlusNous Portal | $0.005 | $0.455 | $0.005 | — | $22 of usage credits · $10 rollover cap | ||
| Portal SuperNous Portal | $0.005 | $0.455 | $0.005 | — | $110 of usage credits · $50 rollover cap | ||
| Portal UltraNous Portal | $0.005 | $0.455 | $0.005 | — | $220 of usage credits · $100 rollover cap | ||
| OpenRouter APIOpenRouter | $0.006 | $0.527 | $0.005 | 108 tps | — | ||
| DevPass LiteLLM Gateway | $0.068 | $0.27 | $0.004 | — | $58 of model usage | ||
| DevPass ProLLM Gateway | $0.068 | $0.27 | $0.004 | — | $158 of model usage | ||
| DevPass MaxLLM Gateway | $0.068 | $0.27 | $0.004 | — | $358 of model usage | ||
| LLM Gateway APILLM Gateway | $0.142 | $0.567 | $0.007 | — | — | ||
| DeepInfra APIDeepInfra | $0.2 | $0.6 | $0.006 | — | — | ||
| Entrim APIEntrim | $0.2 | $0.6 | $0.005 | — | — | ||
Portal Plus
OpenRouter API
DevPass Lite
DeepInfra API
Entrim APIEffective prices apply the gateway's service fee, and scale a subscription's listed rate by its monthly pool — dollars for a dollar-credit plan, credits for a token-credit plan. A subscription's effective rate holds only if you use the full pool. The fee is included only where the gateway publishes it, and sales tax is never included — confirm both with the provider.
Peak endpoints
The rate that applies inside the peak windows.
| Variant | Plan | Pricing /1M | Privacy | Performance | Monthly quota | |||
|---|---|---|---|---|---|---|---|---|
| Input | Output | Cached read | Logs | Trains | ||||
| DeepSeek V4.1 Flash (Peak) | OpenCode GoOpenCode | $0.03 | $0.12 | $0.001 | — | $100 | ||
| DeepSeek V4.1 Flash Latest (Peak) | Command Code GOATCommand Code | $0.03 | $0.12 | $0.001 | — | $100 | ||
| DeepSeek V4.1 Flash Latest (Peak) | Command Code ProCommand Code | $0.055 | $0.218 | $0.001 | — | $110 | ||
| DeepSeek V4.1 Flash (Peak) | OpenCode Go TeamsOpenCode | $0.057 | $0.229 | $0.001 | — | $105 | ||
| DeepSeek V4.1 Flash (Peak) | OpenCode Go PlusOpenCode | $0.06 | $0.24 | $0.001 | — | $200 | ||
| DeepSeek V4.1 Flash (Peak) | Ollama Cloud ProOllama | $0.1 | $0.4 | $0.002 | — | $60 of usage credits | ||
| DeepSeek V4.1 Flash (Peak) | Ollama Cloud MaxOllama | $0.1 | $0.4 | $0.002 | — | $300 of usage credits | ||
| DeepSeek V4.1 Flash (Peak) | OpenRouter APIOpenRouter | $0.158 | $0.633 | $0.016 | 116 tps | — | ||
| DeepSeek V4.1 Flash Latest (Peak) | Command Code Max 10xCommand Code | $0.2 | $0.8 | $0.004 | — | $150 | ||
| DeepSeek V4.1 Flash Latest (Peak) | Command Code Max 20xCommand Code | $0.2 | $0.8 | $0.004 | — | $300 | ||
OpenCode Go
Command Code GOAT
Ollama Cloud ProEffective prices apply the gateway's service fee, and scale a subscription's listed rate by its monthly pool — dollars for a dollar-credit plan, credits for a token-credit plan. A subscription's effective rate holds only if you use the full pool. The fee is included only where the gateway publishes it, and sales tax is never included — confirm both with the provider.
Off-peak endpoints
The rate that applies outside the peak windows.
| Variant | Plan | Pricing /1M | Privacy | Performance | Monthly quota | |||
|---|---|---|---|---|---|---|---|---|
| Input | Output | Cached read | Logs | Trains | ||||
| DeepSeek V4.1 Flash (Off-Peak) | OpenCode GoOpenCode | $0.015 | $0.06 | $0 | — | $100 | ||
| DeepSeek V4.1 Flash Latest (Off-Peak) | Command Code GOATCommand Code | $0.015 | $0.06 | $0 | — | $100 | ||
| DeepSeek V4.1 Flash Latest (Off-Peak) | Command Code ProCommand Code | $0.027 | $0.109 | $0.001 | — | $110 | ||
| DeepSeek V4.1 Flash (Off-Peak) | OpenCode Go TeamsOpenCode | $0.029 | $0.114 | $0.001 | — | $105 | ||
| DeepSeek V4.1 Flash (Off-Peak) | OpenCode Go PlusOpenCode | $0.03 | $0.12 | $0.001 | — | $200 | ||
| DeepSeek V4.1 Flash (Off-Peak) | Ollama Cloud ProOllama | $0.05 | $0.2 | $0.001 | — | $60 of usage credits | ||
| DeepSeek V4.1 Flash (Off-Peak) | Ollama Cloud MaxOllama | $0.05 | $0.2 | $0.001 | — | $300 of usage credits | ||
| DeepSeek V4.1 Flash (Off-Peak) | OpenRouter APIOpenRouter | $0.079 | $0.316 | $0.008 | 116 tps | — | ||
| DeepSeek V4.1 Flash Latest (Off-Peak) | Command Code Max 10xCommand Code | $0.1 | $0.4 | $0.002 | — | $150 | ||
| DeepSeek V4.1 Flash Latest (Off-Peak) | Command Code Max 20xCommand Code | $0.1 | $0.4 | $0.002 | — | $300 | ||
Effective prices apply the gateway's service fee, and scale a subscription's listed rate by its monthly pool — dollars for a dollar-credit plan, credits for a token-credit plan. A subscription's effective rate holds only if you use the full pool. The fee is included only where the gateway publishes it, and sales tax is never included — confirm both with the provider.
Benchmarks
DeepFrugal tracks Intelligence, Coding and Agentic scores for the models it catalogs. Scores appear here when Artificial Analysis publishes them; a model with no published score shows no table. Treat one score as a summary, not a verdict.
How DeepSeek V4.1 Flash is priced
A gateway can bill the same model in more than one shape. Each shape lands on its own row, so you compare like with like:
- Pay as you go. You pay per token on input and output.
- Cached input. Reused prompt tokens usually bill at a lower rate.
- Cache writes. Some providers charge a fee to store a prompt in the cache.
- Long context. Requests above a context threshold can bill at a higher rate.
- Peak and off-peak. A provider that varies its rate publishes a separate row for each window.
- Subscription quota. A monthly plan covers usage up to a published quota.
DeepSeek publishes two windows for this model. Off-peak rates are half the peak rate. Peak hours run 01:00–04:00 and 06:00–10:00 UTC, Monday to Friday; every other hour is off-peak. A gateway that resells the model can then add a service fee; sales tax is not included and depends on your billing country. Each provider sets its own rate for the same weights.
Pay-as-you-go or subscription?
A subscription only wins above a usage threshold. Below it, you pay for capacity you never use. The threshold depends on your monthly mix: how much input, cached read, cache write and output you send.
The break-even guide explains how DeepFrugal turns a plan's list price into an effective rate per token, and how AI subscription credits work explains what the quota covers. The break-even calculator runs the same maths on your own usage.
Best plan for this model
Providers split into two groups. Some publish a peak and an off-peak rate; others publish one flat rate for the model. The two finders below rank each group on the same monthly workload (a mix of input, cached read, cache write and output). Both value the plans on this model's usage, not on the listed token rate alone.
Best plan with peak and off-peak rates
The first finder splits the workload across the two windows: peak usage is one fifth of off-peak usage. It ranks the plans that publish both rates.
Best plan with a flat rate
The second finder bills the whole workload at one flat rate. It ranks the providers that make no distinction between windows and publish one variant.
Change the usage numbers to your own to see how a ranking moves. Rank by reorders the list: cheapest, best privacy, performance, or the broadest catalog. Open How the best plan is calculated for the method.
What to check beyond price
Two endpoints with the same rate are still not equal. Check:
- Context window. A long-context tier can cost more and cap at a lower size.
- Throughput. Tokens per second tells you whether the endpoint keeps up with an interactive or agentic workload.
- Logging and training. The main table carries a privacy column per row. Treat a blank as unknown, never as "no".
- Rate limits. A subscription can throttle bursts well below its monthly quota.
Summary
The cheapest way to run DeepSeek V4.1 Flash depends on your usage, not on the list price alone. Compare the endpoints first, then value your own month against the plans. Remember that the rate depends on the time of day.
Next steps
- Rank every plan on your own DeepSeek V4.1 Flash usage in the plan finder.
- Compare two subscriptions on this model in the plan comparison calculator or the plan head-to-head.
- Check whether a plan beats pay-as-you-go in the break-even calculator.
- Filter the live table by context, latency and privacy.
Frequently asked questions
Where is DeepSeek V4.1 Flash cheapest?
The cheapest-endpoint widget ranks every gateway by listed input rate, with a toggle for effective prices.
Is a subscription cheaper than an API for this model?
It depends on usage. The plan finder widget ranks quota plans for a monthly mix, and the break-even calculator shows where the plan wins.
What changes between peak and off-peak?
The listed rate. DeepFrugal tracks peak and off-peak as separate variants, each with its own window.
Do the shown prices include fees and tax?
The effective-prices toggle applies a gateway's service fee; sales tax is never included. Confirm it with the provider.
Why does the same model cost more on one gateway?
A gateway marks up the underlying provider and can add a service fee and sales tax. Each step raises the effective rate.
How do peak hours change the bill?
Usage inside the peak windows bills at the higher rate; usage outside them bills at half. A workload that runs mostly off-peak reaches a lower effective rate than one that runs at peak, on the same plan.
Can cached reads change the ranking?
Yes. A cache-heavy workload shifts the ranking, because a cache read often costs a fraction of fresh input. Enter your own cached read and cache write volumes in the calculators to see the effect.
Does a subscription make DeepSeek V4.1 Flash cheaper?
Only above a usage threshold. Under it, pay-as-you-go is cheaper. The threshold moves with your mix, because cached reads and output bill at different rates.