LLM Pricing
ModelsLeaderboardsToolsProvidersReleasesArticles

© 2026 LLM Pricing

About
·Contact
·Privacy
·RSS
← Model list

DeepSeek V4 Flash

deepseek·deepseek/deepseek-v4-flash·GA·Open weights·deepseek-flash series
Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work

Specs & pricing

Input / output per 1M tokens
Official price
—
Lowest paid·UnoRouterGateway
$0.063 / $0.13
Blended $0.078 · 8× spread
5 more $0 channels
Context
1,000,000
Output limit
384,000
Knowledge cutoff
2025-05
Released / updated
2026-04-24 / 2026-04-24
Capabilities
✓ Reasoning✓ Tool use✓ Structured output✓ TemperatureAttachments
⚠ Providers report capability flags inconsistently
Modalities
TextImage
Weights
Hugging Face

DeepSeek V4 Flash is offered by 72 providers on this page. Public prices are shown for 60 of them. No official API price is listed for it; the prices below come from third-party channels. The lowest paid channel is UnoRouter at $0.063 / $0.125 per 1M. That channel is a third-party gateway, so confirm its availability and rate limits before depending on it. Prices vary about 8.0× across the providers that host it. It also has 5 free ($0) channels and 4 covered by a paid subscription; free tiers usually carry rate limits, and subscription-covered access bills $0 per token only after the subscription fee.

The context window is 1,000,000 tokens at the reference host, but hosts report different limits, from 1,000,000 to 1,050,000, so the usable window depends on the provider you pick. It supports reasoning, tool use, and structured output. Accepted input modalities are Text and Image. The weights are open, so it can also be self-hosted or served through a gateway of your choice. Providers report the capability flags inconsistently, so verify a specific feature against the host you plan to use. Its training knowledge cuts off at 2025-05.

Available at 72 providers60 with public prices · 5 free · 4 subscription-covered

  • ProviderInput / Output
  • PendraGateway
    Free
    Context 1,000,000Output limit 384,000
  • SCNet Token PlanGateway
    DeepSeek-V4-Flash
    Subscription
    Cache read $0Context 1,000,000Output limit 384,000
  • Alibaba Token PlanGateway
    Subscription
    Cache read $0Context 1,000,000Output limit 384,000deprecated
  • Alibaba Token Plan (China)Gateway
    Subscription
    Cache read $0Context 1,000,000Output limit 384,000deprecated
  • InferXGateway
    Free
    Context 1,000,000Output limit 100,000
  • UnoRouterGateway
    deepseek-v4-flash:free
    Free
    Context 1,000,000Output limit 384,000deprecated
  • SenseNova (China)Gateway
    Free
    Cache read $0Context 1,000,000Output limit 65,536
  • OrcaRouterGateway
    deepseek/deepseek-v4-flash-free
    Free
    Context 1,000,000Output limit 384,000
  • Volcengine Ark Coding PlanGateway
    Subscription
    Cache read $0Context 1,000,000Output limit 384,000
  • UnoRouterGateway
    $0.063 / $0.13
    Context 1,000,000Output limit 384,000deprecated
ProviderTierInputOutputCache readCache writeContextOutput limitStatus
PendraGatewayFree——1,000,000384,000
SCNet Token Plan
DeepSeek-V4-Flash
GatewaySubscription$0—1,000,000384,000
Alibaba Token PlanGatewaySubscription$0$01,000,000384,000deprecated
Alibaba Token Plan (China)GatewaySubscription$0$01,000,000384,000deprecated
InferXGatewayFree——1,000,000100,000
UnoRouter
deepseek-v4-flash:free
GatewayFree——1,000,000384,000deprecated
SenseNova (China)GatewayFree$0—1,000,00065,536
OrcaRouter
deepseek/deepseek-v4-flash-free
GatewayFree——1,000,000384,000
Volcengine Ark Coding PlanGatewaySubscription$0—1,000,000384,000
UnoRouterGateway$0.063$0.13——1,000,000384,000deprecated

Sorted by blended price (input×0.75 + output×0.25) asc. The official channel always shows regardless of rank. Whether a gateway's low price is actually usable can't be verified.

Reasoning control

effort = loweffort = higheffort = maxToggle (on / off)effort = noneeffort = xhigheffort = mediumeffort = minimalbudget_tokens ≥ 128 ≤ 32,768

Interleaved thinking (reasoning between tool calls) is declared by 43 of 72 providers.

12 / 72 providers expose no reasoning control (reasoning_options: []).

Your usage cost

Common scenarios
1LLM Gateway$15.41
2DevPass (LLM Gateway)$15.41
3Standard Compute$16.85
4LLM Gateway$18.36
5OpenRouter$18.36
6Kilo Gateway$18.36

9 more channels offer $0 (Pendra, SCNet Token Plan, Alibaba Token Plan etc.); free tiers usually have rate limits and no SLA, excluded from ranking.

The cheapest paid channel is the only channel.

Price historyone sample accumulated per data sync

Input listOutput listMin blended
$0.60$02026-08-052026-10-16
2026-10-16Input list —Output list —Min blended $0.078

Benchmark22 items

NameConditionsScoreMetricSource
SWE-Bench Verified—79resolvedSource ↗
MMLU-Provariant: preview checkpoint; max effort86.2EMSource ↗
SimpleQA-Verifiedvariant: preview checkpoint; max effort34.1pass@1Source ↗
Chinese SimpleQAvariant: preview checkpoint; max effort78.9pass@1Source ↗
GPQA Diamondvariant: preview checkpoint; max effort88.1pass@1Source ↗
Humanity's Last Examvariant: preview checkpoint; max effort; without tools34.8pass@1Source ↗
LiveCodeBenchvariant: preview checkpoint; max effort91.6pass@1Source ↗
Codeforcesvariant: preview checkpoint; max effort3052ratingSource ↗
HMMTvariant: preview checkpoint; max effort · dataset: February 202694.8pass@1Source ↗
IMOAnswerBenchvariant: preview checkpoint; max effort88.4pass@1Source ↗
MathArena Apexvariant: preview checkpoint; max effort33pass@1Source ↗
MathArena Apex Shortlistvariant: preview checkpoint; max effort85.7pass@1Source ↗
MRCRvariant: preview checkpoint; max effort · dataset: 1M context78.7MMRSource ↗
CorpusQAvariant: preview checkpoint; max effort · dataset: 1M context60.5accuracySource ↗
Terminal-Benchvariant: preview checkpoint; max effort · v2.056.9accuracySource ↗
SWE-Bench Provariant: preview checkpoint; max effort52.6resolvedSource ↗
SWE-Bench Multilingualvariant: preview checkpoint; max effort73.3resolvedSource ↗
BrowseCompvariant: preview checkpoint; max effort73.2pass@1Source ↗
Humanity's Last Examvariant: preview checkpoint; max effort; with tools45.1pass@1Source ↗
MCP Atlasvariant: preview checkpoint; max effort · dataset: public69pass@1Source ↗
GDPval-AAvariant: preview checkpoint; max effort1395EloSource ↗
Toolathlonvariant: preview checkpoint; max effort47.8pass@1Source ↗

The same benchmark scores very differently across harness / dataset, so the qualifying conditions must be shown together.

Related models

DeepSeek V4.1 Flash 1017same series$0.15 / $0.60DeepSeek: DeepSeek Flash Latestsame series—DeepSeek V4.1 Flashsame series—DeepSeek V3.2 Expsame lab—DeepSeek V4 Flash 0423same lab—
Data partly from models.dev (MIT)