AI Rate Card

DeepSeek-V3.1 vs DeepSeek V4.1 Flash: API price and specs

DeepSeek V4.1 Flash is 76% cheaper than DeepSeek-V3.1 on input price.

DeepSeek V4.1 Flash is 79% cheaper than DeepSeek-V3.1 on output price.

DeepSeek V4.1 Flash is 77% cheaper than DeepSeek-V3.1 on a 1M-in / 200k-out workload.

DeepSeek V4.1 Flash has a 539.4% larger context than DeepSeek-V3.1.

FieldDeepSeek-V3.1DeepSeek V4.1 Flash
Lowest input, per 1M$0.19 +322.2%$0.045
Median input, per 1M$0.27$0.30 +12.4%
Lowest output, per 1M$0.71 +373.3%$0.15
Median output, per 1M$1.00$1.20 +20%
Lowest unit price——
Cost of 1M in + 200k out$0.33 +342.7%$0.075
Context164K +84.4%1M
Max output164K +84.4%1M
Providers18 +70%60
RegionsUSAPAC, EU, GLOBAL, US
Acceptstextapplication/pdf, file, image, image/gif, image/jpeg, image/png, image/webp, pdf, text
Returnstexttext
TypeText modelsText models
Free optionnoyes
Cheapest providermeganovaopenrouter

Your own workload: how many tokens the model gets (input) and gives back (output) - for example, in a month. The "Cost" row of the table above is recalculated for these numbers, for each model.

Computed from current offers each time the page opens; the best value of a row is bold, the percent is how far behind the best.

Providers selling both

aihubmix, aiml, baseten, deepinfra, edenai, fastrouter, google-vertex, huggingface, jiekou, meganova, poe, requesty, siliconflow, togetherai, trustedrouter, vercel

Frequently asked questions

Which is cheaper, DeepSeek-V3.1 or DeepSeek V4.1 Flash?

DeepSeek V4.1 Flash is 76% cheaper than DeepSeek-V3.1 on input price. DeepSeek V4.1 Flash is 79% cheaper than DeepSeek-V3.1 on output price. DeepSeek V4.1 Flash is 77% cheaper than DeepSeek-V3.1 on a 1M-in / 200k-out workload.

Which has the larger context, DeepSeek-V3.1 or DeepSeek V4.1 Flash?

DeepSeek-V3.1: 164K tokens; DeepSeek V4.1 Flash: 1M tokens.

Where can I use both DeepSeek-V3.1 and DeepSeek V4.1 Flash?

aihubmix, aiml, baseten, deepinfra, edenai, fastrouter, google-vertex, huggingface, jiekou, meganova, poe, requesty, siliconflow, togetherai, trustedrouter, vercel