Appndo
All models

DeepSeek V3

DeepSeek164K contextTool calling

LMArena score 1332 · #223 overall · #236 in coding (2026-10-13)

Official price from DeepSeek

Per million tokens

Input
$0.27
Output
$1.10
Cached input
$0.07
Batch input / output
—
  • Writing to the prompt cache costs $0 per million tokens.
  • Up to 8K output tokens per request.

Estimate your monthly cost

Est. per month = input tokens × input price + output tokens × output price. Prices are per million tokens, so 10M input tokens at $3.00 cost $30.

From DeepSeek: $4.90 a month. Hyperbolic is the cheapest seller at $2.40, $2.50 less.

Where to buy it (9)

Sorted by your monthly estimate. Prices per million tokens.

  • HyperbolicCheapest

    Input / output
    $0.20 / $0.20
    Context
    33K
    Per month
    $2.40

    hyperbolic/deepseek-ai/DeepSeek-V3

  • DeepSeekOfficial

    Input / output
    $0.27 / $1.10
    Context
    66K
    Per month
    $4.90

    deepseek/deepseek-v3

  • DeepInfra

    Input / output
    $0.32 / $0.89
    Context
    164K
    Per month
    $4.98

    deepinfra/deepseek-ai/DeepSeek-V3

  • Nebius

    Input / output
    $0.50 / $1.50
    Context
    128K
    Per month
    $8.00

    nebius/deepseek-ai/DeepSeek-V3

  • Novita AI

    Input / output
    $0.89 / $0.89
    Context
    64K
    Per month
    $10.68

    novita/deepseek/deepseek_v3

  • Fireworks AI

    Input / output
    $0.90 / $0.90
    Context
    131K
    Per month
    $10.80

    fireworks_ai/accounts/fireworks/models/deepseek-v3

  • Vercel AI Gateway

    Input / output
    $0.90 / $0.90
    Context
    128K
    Per month
    $10.80

    vercel_ai_gateway/deepseek/deepseek-v3

  • Together AI

    Input / output
    $1.25 / $1.25
    Context
    66K
    Per month
    $15.00

    together_ai/deepseek-ai/DeepSeek-V3

  • Replicate

    Input / output
    $1.45 / $1.45
    Context
    66K
    Per month
    $17.40

    replicate/deepseek-ai/deepseek-v3

Cloud and hosted sellers may add regional pricing, minimums or platform fees. Supports: Tool calling.

Similar models