Home › Blog › DeepSeek API Pricing

DeepSeek API Pricing in 2026: What the Peak/Off-Peak Change Means for You

In August 2026 DeepSeek replaced flat pricing with a peak/off-peak model — peak-hour input tripled to ¥9 per 1M tokens. Here's the full comparison and how Bridge keeps DeepSeek V4 Pro affordable around the clock.

The change: peak/off-peak pricing (effective August 2026)

Alongside the V4-Pro-0813 release, DeepSeek introduced time-of-day pricing. On weekdays, peak hours — 9:00–12:00 and 14:00–18:00 Beijing time — cost double. Weekends are billed entirely off-peak. This replaced the flat ¥3 input / ¥6 output pricing announced in May 2026.

Bridge pricing: flat, all day, every day

Bridge prices are flat — no peak hours to schedule around — and come from a live price table you can audit on the models page. V4 Flash is available for high-volume customers at $0.12/$0.32.

Why cache hits change the math

DeepSeek's context caching charges cache-hit input at a fraction of the miss price. On Bridge the hit price is $0.025 per 1M tokens — a 10x discount over the miss price. Long multi-turn conversations and agent loops with stable system prompts see the biggest savings; Bridge meters cached tokens automatically and shows them on every request receipt.

Code: swap in two lines

# Python (openai SDK) — change base_url and api_key, keep everything else
from openai import OpenAI

client = OpenAI(
    base_url="https://bridge.computeindex.net/v1",
    api_key="sk-bridge-YOUR_KEY",
)

completion = client.chat.completions.create(
    model="deepseek-v4-pro",
    messages=[{"role": "user", "content": "Explain peak/off-peak pricing."}],
)
print(completion.choices[0].message.content)

FAQ

Is DeepSeek V4 Pro on Bridge the same model as official? Yes — same model id family, served through Bridge's OpenAI-compatible endpoint with per-token metering and cached-input billing.

Does Bridge have peak pricing? No. Prices are flat and listed on the live price table. If costs change, the price table changes — you always see current pricing before you call.

Which DeepSeek models are available? DeepSeek V4 Pro (flagship) and V4 Flash (high-volume tier).