DeepSeek · Premium Reasoning

DeepSeek V4 Pro

deepseek/deepseek-v4-pro DeepSeek's premium reasoning model, accessible through Leanroute's OpenAI-compatible AI Gateway. Pay provider list price with zero per-request markup; flat $15/mo BYOK subscription covers the platform fee.

Pricing snapshot: August 2026. Re-validated monthly against the upstream DeepSeek pricing page.

Pricing

TypeOff-peakPeak (2×)
Input tokens$0.435$0.870
Cached input (99% off)$0.0044$0.0087
Output tokens$0.870$1.74
Peak-hour surcharge. DeepSeek charges 2× the listed rate during 01–04 + 06–10 UTC. Leanroute's router treats DeepSeek as 2× more expensive during those windows — when cheaper-model routing is on, requests automatically shift to a same-tier alternative outside your peak window. Response header x-gateway-time-tier: peak tells you when a request landed in the peak window.

Leanroute's markup: $0. You pay the provider list price above, plus a flat $15/mo BYOK subscription (or 5% at top-up for prepaid credits). Compare pricing at /calculator.

Capabilities

Premium Reasoning tierGlobal (US-incorporated)StreamingTool calls

Example request

curl https://api.leanroute.dev/v1/chat/completions \
  -H "Authorization: Bearer $LEANROUTE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek/deepseek-v4-pro",
    "messages": [
      { "role": "user", "content": "Hello, DeepSeek V4 Pro!" }
    ]
  }'

OpenAI-wire-compatible — point any OpenAI SDK at api.leanroute.dev/v1 and set model: "deepseek/deepseek-v4-pro". Full endpoint reference at /docs/api.

Other DeepSeek models

Same-tier alternatives across providers

Leanroute's router picks the cheapest same-tier model automatically when cheaper-model routing is on. Config at /dashboard/settings.

See also: Full model catalog · API reference · Savings calculator · Leanroute pricing