KymaKyma
DeepSeek

DeepSeek

DeepSeek V4 Flash

#1 on KymaBest value

DeepSeek's fast-and-cheap V4-tier model with a 1M-token context window. Reach for it when you want strong quality at value pricing for coding, general tasks, and long-context work.

Modalities

Text → Text

Below release

None reported

Weights

Published

deepseek-ai/DeepSeek-V4-Flash

Where it sits in the catalogue Kyma measures

Every number against every text model Kyma prices per token — a stated rule, not a chosen line-up.

Price · input#8/70

$0.1389

output $0.2778 · cached $0.0243

median $0.977 · best $0.0494

Price · output#4/70

$0.2778

per 1M tokens generated

median $3.4515 · best $0.2164

Availability · 30d#3/70

99.93%

48,418 observations since 2026-05-02

median 93.8% · best 100.0%

Throughput#56/70

27.5 tok/s

probe median, one fixed prompt

median 39.3 tok/s · best 313.6 tok/s

Response time#16/70

1.41 s

probe median, to a complete answer

median 2.45 s · best 0.84 s

Context

1M

max output 65,536

published by DeepSeek, not measured here

Tick above each rail is this model, below it the other 69. Dashed rule is the field median, solid is its best. Better is left; the axis stops at the 90th percentile, so a few models sit past its right edge.

Usage

How much this model is actually called here.

Rank

#1

of 117 active models

Tokens served

1.1B

all-time

Platform share

32.4%

of all tokens

Tokens · last 15 daysAug 27 → Sep 10

How it behaves under real clients

The same model answering different prompt shapes — measured, not benchmarked.

ClientRequestsTokensThroughputTo first tokenp95 totalCompleted
OpenAI Python45,374625.4M26 tok/s4.42 s12 s100%
Python App1,1134.1M43 tok/s9.12 s69 s99.9%
Node.js App4275.4M42 tok/s60 s295 s97.9%
OpenClaw1356.0M17 tok/s3.38 s27 s100%

Grouped by the client that sent the request. Not a benchmark: the same model on different prompt shapes. Throughput and time-to-first-token are medians; p95 is the slowest response in twenty. 2 clients under 100 requests not shown.

Two clocks, and why they disagree

Kyma measures this model twice. Both are real; they answer different questions.

Probe · every 6h · 30 days

1.41s to answer

One fixed prompt, on a schedule, to every model. Comparable, because the model is the only thing that changes.

Observations
48,418
Answered by a substitute
35

Real traffic · last 7 days

6.11s to answer

Your requests, at the lengths clients actually send. Not comparable between models, but it is what running this one feels like.

Requests
1,202
Completed
99.6%
p95
76.3 s

Caching, as realised · last 7 days

68.2% of input cached

The share of input that actually hit cache, so the effective rate below is what was charged, not a best case.

Cached input
24.9M
Fresh input
11.6M
List input
$0.1389 /1M
Effective input
$0.060694 /1M

The gap is prompt length, not the model degrading. Use the probe figure to choose between models, the traffic figure to budget for your own.

Pricing

Pay per token. Cached input bills at this model’s own cached rate, listed below.

$0.1389 /1M input$0.2778 /1M output
Hobby10 req/day · 2K in / 500 out
~$0.13/mo
Production1,000 req/day · 2K in / 500 out
~$12.50/mo
Scale20,000 req/day · 2K in / 500 out
~$250/mo
+ Estimate your workload
1,000
2,000
500
30%

Estimated monthly cost

$10.44

$0.3479 / day on DeepSeek V4 Flash

Same workload on:

DeepSeek V4 Flash Vision$31.18+199%
DeepSeek V4 Pro$62.11+495%

Estimates use list pricing with cached input at this model's own cached rate. Actual bills depend on real token counts, and every response includes its exact cost.

When to use DeepSeek V4 Flash

Updated 2026-06-10

Where this model earns its cost — and where it doesn't.

DeepSeek V4 Flash is the fast, cheap entry in DeepSeek's V4 tier — a 284B mixture-of-experts model that sits in Kyma's strong quality tier. It's built as a value workhorse: general tasks, coding, and long-context jobs where you want most of the quality at a fraction of the cost.

Coding agents are its biggest consumers by a wide margin. Every call gets Kyma's automatic failover, and prompt caching is fully supported, so repeated prompt prefixes bill at this model's cached input rate.

Capability-wise it checks every box for agent work: a 1M-token context window, up to 65K tokens of output per request, function calling, structured outputs, and extended reasoning. Input and output are text-only.

Coding

A core recommended use, and where its production traffic goes — OpenClaw, Roo Code, and Cline are among its top apps on Kyma, with OpenClaw alone at roughly half its tokens.

Long-context analysis

The 1M-token window takes entire codebases, document collections, or long agent transcripts in one request.

High-volume pipelines

Cheap-tier pricing makes it a sensible default for background jobs, batch processing, and anything that runs thousands of times a day.

Structured agent backbones

Function calling plus structured outputs keep multi-step agents and extraction pipelines on schema.

Reasoning on a budget

Extended reasoning support handles planning and analysis steps without paying flagship rates.

Not ideal for: Image inputs — it's text-only, so anything that needs to look at screenshots or photos should go to a vision model instead.

How it compares

Against the peers people actually weigh it against.

SpecDeepSeek V4 FlashDeepSeek V4 Flash VisionDeepSeek V4 Pro
Input /1M$0.1389$0.297$0.6901
Output /1M$0.2778$0.891$1.38
Context1M1M1M
ToolsYesYesYes
ReasoningYesYesYes
Throughput27.5 tok/s81.1 tok/s31.3 tok/s

Quick start

Up and running in under two minutes.

  1. 1

    Create an API key

    Sign up and grab a key from the dashboard — $0.50 free credit on the free tier, which covers DeepSeek V4 Flash. No card required.

    Get API key →
  2. 2

    Make your first request

    Drop in your key and send a chat completion — fully OpenAI-compatible.

    curl https://kymaapi.com/v1/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-v4-flash",
    "messages": [
      {"role": "user", "content": "Explain prompt caching in one paragraph."}
    ]
  }'
  3. 3

    Stream responses

    Add "stream": true to receive tokens as they arrive.

    curl https://kymaapi.com/v1/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-v4-flash",
    "stream": true,
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

FAQ

Common questions about this model.

What is the context window of DeepSeek V4 Flash?

DeepSeek V4 Flash has a 1M-token context window — roughly 1471 pages of text in a single request.

How much does the DeepSeek V4 Flash API cost?

$0.1389 per 1M input tokens and $0.2778 per 1M output tokens, with cached input at $0.0243/1M — 83% off the input rate on repeated prompt prefixes. No subscription; you pay only for what you use.

Does DeepSeek V4 Flash support function calling?

Yes — DeepSeek V4 Flash supports tool/function calling and structured outputs (JSON mode), so it works with agent frameworks out of the box.

Are the weights for DeepSeek V4 Flash publicly available?

Yes. DeepSeek publishes DeepSeek V4 Flash's weights as deepseek-ai/DeepSeek-V4-Flash, so you can download and run the model yourself (https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash, read 2026-08-14). Kyma serves it because it is convenient and has failover behind it, not because it is the only way to reach it.

Is DeepSeek V4 Flash ever served below the precision its creator released it at?

Not as far as Kyma can see. DeepSeek V4 Flash was released by DeepSeek at fp4 + fp8 (https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash, read 2026-08-14), and no route serving it here reports a format below that, as checked on 2026-09-03. Read that as the absence of a claim rather than as a guarantee: a route publishes its precision only where it chooses to, and most state nothing either way.

How do I use DeepSeek V4 Flash?

Kyma is OpenAI-compatible: point your SDK's base URL at https://kymaapi.com/v1, use your Kyma API key, and set the model to deepseek-v4-flash. Signing up is free and includes $0.50 of credit on the free tier, which covers this model — no card required.

Is DeepSeek V4 Flash good for coding agents?

Yes — coding is one of its recommended uses, and Kyma's production data backs it up: OpenClaw, Roo Code, and Cline are among its top apps by tokens. Function calling, structured outputs, and the 1M-token context let an agent hold a large codebase in view while making tool calls reliably.

When should I pick V4 Flash over a flagship-tier model?

When volume or context size matters more than squeezing out the last bit of quality. It sits in the strong quality tier at cheap-tier pricing, so it's the better trade for high-frequency agent loops, batch jobs, and long-context work. If a task is a one-shot where peak output quality is everything, a flagship-tier model is worth the premium.

Why use DeepSeek V4 Flash through Kyma?

One API key and one OpenAI-compatible endpoint cover this and every other model on the platform. You get automatic failover when a serving path degrades, prompt caching on repeated prefixes, and the exact cost of every request in usage.cost.

Start with $0.50 free credit on the free tier — no card required.Create account →

More models by DeepSeek

ModelContextInputOutput
DeepSeekDeepSeek V4 Flash Vision1M$0.297$0.891
DeepSeekDeepSeek V4 Pro1M$0.6901$1.38
DeepSeekDeepSeek V3160K$0.351$0.513
DeepSeekDeepSeek R164K$0.7425$2.957