DeepInfra is the cheapest way to call DeepSeek V4.1 Flash — from $0.31 per combined million tokens
Asked:
“where is the best value (cost per token) to use "DeepSeek V4.1 Flash" as a api”
This compares 9 current provider/tier price points for DeepSeek V4.1 Flash across four independent hosts — DeepInfra, SiliconFlow, Parasail and Tiyuvta — checked 2026-10-21. All prices are USD per 1M tokens; bars rank each option by the cost of a workload of 1M input plus 1M output tokens. DeepInfra lists both the latest 1017 revision and the original endpoint; its prices are promotional with the discount already applied.
Cost for 1M input + 1M output tokens, cheapest first
input $/1M
output $/1M
promotional price, discount already applied
solid labels = DeepSeek-V4.1-Flash-1017 revision; grey name = original V4.1 Flash endpoint
DeepInfra Flex on the 1017 revision is the absolute cheapest: $0.0784 input, $0.2352 output and $0.0024 cached input per 1M — $0.3136 for 1M of each.
deepinfra.comDeepInfra Standard on 1017 ($0.098/$0.294, $0.392 combined) is the safer default when latency and availability matter; both are promotional at 51% off already applied, so the ranking may change when the promotion ends.
deepinfra.comNeed the original V4.1 Flash endpoint rather than the 1017 update? DeepInfra Flex is $0.112/$0.336 and Standard $0.14/$0.42 per 1M, and SiliconFlow ($0.15/$0.60, $0.003 cached) is the strongest independent alternative.
siliconflow.comParasail and Tiyuvta both charge $0.30 input / $1.20 output — $1.50 combined, nearly 5× the cheapest option.
parasail.io
All 9 price points
| Provider | Tier | Model revision | Input $/1M | Output $/1M | Cached in $/1M | 1M in+1M out | Conditions | Source |
|---|
Data: 9 provider/tier pricing rows for DeepSeek V4.1 Flash from four source hosts, checked 2026-10-21. Prices are USD per million tokens; the combined column is 1M input + 1M output. All rows are shown; nothing was cut. DeepInfra prices are promotional with the stated discount already applied.