DeepSeek V4 Pricing Update — Peak/Off-Peak Rates Effective August 17, 2026

DeepSeek introduced peak/off-peak pricing on August 17, 2026 (effective immediately for all V4 series models). This is the first upstream-tier change to the V4 Flash / V4 Pro pricing structure since launch.

We've adjusted our advertised rates to match — here's everything you need to know.

What Changed

DeepSeek V4 Flash and V4 Pro now use peak/off-peak pricing:

| Model | Previous Output | Peak (weekdays 9:00–12:00, 14:00–18:00 CST) | Off-peak | |-------|-----------------|---------------------------------------------|----------| | DeepSeek V4 Flash | ¥2/M | ¥9/M (4.5×) | ¥3–5/M | | DeepSeek V4 Pro | ¥6/M | ¥27/M (4.5×) | ¥9–15/M |

Peak hours are weekdays 9:00–12:00 and 14:00–18:00 China Standard Time (UTC+8). Off-peak is nights, weekends, and Chinese holidays.

What We Did at Global API

We adjusted our advertised pricing to maintain a consistent margin (~25%) across both peak and off-peak windows:

| Model | Our Backend Cost (1 credit = $0.01) | Per Million Tokens | |-------|-------------------------------------|---------------------| | DeepSeek V4 Flash | 26 input / 90 output credits | $0.26 input / $0.90 output | | DeepSeek V4 Pro | 80 input / 270 output credits | $0.80 input / $2.70 output |

Compared to direct DeepSeek peak pricing, Global API remains ~28% cheaper even at peak hours:

• V4 Flash: $0.90/M output vs DeepSeek's $1.25/M peak
• V4 Pro: $2.70/M output vs DeepSeek's $3.75/M peak

What This Means for Your Existing Credits

Nothing changes for unused credits. Your credit balance retains its purchasing power based on the prices you saw when you purchased it. This update only affects future credit consumption.

If you bought a Pro Pack before August 17, 2026 at the old rate, those credits still deduct at the original model rates — the new rates apply to credits purchased after this update takes effect on our platform.

How to Optimize Costs

Three practical patterns to minimize the impact of peak pricing:

1. Batch background work to off-peak — Anything that doesn't need real-time response (data processing, RAG indexing, bulk summarization) should run between 18:00 and 9:00 CST or on weekends. DeepSeek's off-peak rate is roughly half of peak.

2. Use GA Economy for background tasks — Our GA Economy route (deepseek-v4-flash based) automatically benefits from the same 25% margin we maintain. Same cost profile, fewer integration headaches.

3. Reserve V4 Pro for hard reasoning tasks — V4 Pro's peak price jump is the steepest (4.5×). Use V4 Flash for everything except multi-step reasoning, code review, and long-context work where Pro's quality justifies the cost.

FAQ

Q: Are my existing credits affected? No. Credit deducts at the rate that applied when you purchased. Only credits purchased after this update use the new rate.

Q: Will you change pricing again? If upstream pricing changes, yes — we follow DeepSeek's rates with a stable margin. We'll always announce in advance here.

Q: What about GA Standard / GA Express? GA Economy stays tied to V4 Flash's new rate. GA Standard and Express are V4 Pro tiered (2× and 3× V4 Flash output respectively) and adjusted proportionally.

Q: Can I still get a fixed rate? Yes — the Pro Pack, Business Pack, and Scale Pack all give you more credits per dollar (the larger the pack, the cheaper per million tokens). The price-per-token in your pack is locked in at purchase.

Questions?

Email support@global-apis.com or check our pricing page for live rates.

— Global API Team, August 20, 2026