DeepSeek V4 Flash API Pricing & Cost Calculator

By TechCompare · Updated · How we check

A half-cached 100K-input, 5K-output request costs about $0.0063 at the cheapest listing, DeepInfra, and other hosts price it differently.

A newer version of DeepSeek V4 Flash 0423 is out: DeepSeek V4.1 Flash 1017. Price DeepSeek V4.1 Flash 1017 in the calculator or compare every DeepSeek model.

Calculator

Cost Comparison

Based on 100,000 input tokens (50% cached), 5,000 output tokens, and 1,000 requests.Prices come live from OpenRouter at each maker's own price where it's listed. Temporary deals are left out.

Loading model prices

DeepSeek V4 Flash, released in April 2026, is sold by third-party hosts, and OpenRouter lists no DeepSeek-run endpoint for it. The chart starts from the cheapest host for this request and names it. What you pay depends on the host and its quantization. This page keeps the original V4 Flash URL and prices that model, not the later V4.1 Flash.

DeepSeek: DeepSeek V4 Flash 0423

Cheapest host: DeepInfra (fp8)

Input
$0.09
Output
$0.18
Cached
$0.018

per 1M tokens

Flex available

Price checked 2026-11-19

How this is calculated

The cheapest host for this request, DeepInfra running fp8 weights, lists $0.09 per million uncached input tokens, $0.018 for cache-hit input, and $0.18 for output. For 100,000 input tokens and 5,000 output tokens, a request at those rates costs $0.0099 without cache hits and $0.0063 with half the input cached.

The chart's 1,000 monthly requests therefore cost $6.30. Hosts differ in quantization and maximum output, so compare those before choosing by price alone. Most hosts offer a 1M-token context window.

Verdict

At DeepInfra's rates, a 100K-input, 5K-output request comes to about $0.0099 with no cache hits and $0.0063 when half the input is cached. Over the chart's 1,000 monthly requests that's about $6.30. Compare the host's quantization, output limit, and uptime along with its price, and budget from the actual billed usage. If you're starting a new integration, compare it with DeepSeek V4.1 Flash, which is priced separately.

More API Standalones scenarios

Frequently asked questions

What do hosts charge for DeepSeek V4 Flash?
For this page's request the cheapest listing is DeepInfra, at $0.09 input, $0.018 cache-hit input, and $0.18 output per million tokens. Other hosts set their own rates. Use the rates of the host you actually call, applied to each request's billed usage.
How deep is the cache discount on V4 Flash?
At DeepInfra's $0.09 input and $0.018 cache-hit input, a cache hit is 80% below uncached input. Other hosts discount more or less. Apply the lower price only to tokens reported as cache hits.
Is V4 Flash the same model as V4.1 Flash?
No. V4.1 Flash is a later release with its own listing and prices. This page covers the original V4 Flash, and the V4.1 Flash page covers the newer model. Compare both if you are choosing a model for a new integration.