---
format: "aidr-story-markdown/v1"
id: "2382d0ce57d530dad8289306b4bdb47c88e6635028f5d15e37ae59ec066c97c8"
canonical_url: "https://aidr.today/2382d0ce?lang=en"
title: "DeepSeek Slashes Weekend API Costs to Off Peak Rates From Aug 23 Reversing Aug 16 Hikes"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-08-22T16:49:23.000Z"
category: "Infra"
topics: ["deepseek","inference","agent","llm"]
source_urls: ["https://huggingnews.com/ai/update-deepseek-slashes-weekend-api-costs-to-off-peak-rates-from-aug-23-839a0c76","https://x.com/MegaBlackLabel/status/2091154621154869248","https://x.com/MaxForAI/status/2091150619533009361","https://x.com/hqmank/status/2091160923457352127"]
summary: "Billing for artificial intelligence token usage on Saturdays and Sundays will shift to a standardized low-cost tier to facilitate large-scale computing tasks. DeepSeek is eliminating peak pricing windows for the weekend effective 00:00 Beijing Time on August 23, allowing developers to run batch processing and AI agents at reduced rates. The adjustment aims to optimize the company's compute capacity during traditionally idle periods. The change follows an August 16 update that raised peak rates fivefold and off-peak rates 2.5 times for the V4 Flash model. Daily token usage for that model had previously surged from 3 trillion to 18 trillion tokens on OpenCode within two weeks of its August 1 debut before the price increase stalled growth."
---

# DeepSeek Slashes Weekend API Costs to Off Peak Rates From Aug 23 Reversing Aug 16 Hikes

> [Open the canonical story](<https://aidr.today/2382d0ce?lang=en>)

**Published:** 2026-08-22T16:49:23.000Z
**Category:** Infra
**Topics:** deepseek, inference, agent, llm

## Summary

Billing for artificial intelligence token usage on Saturdays and Sundays will shift to a standardized low\-cost tier to facilitate large\-scale computing tasks\. DeepSeek is eliminating peak pricing windows for the weekend effective 00:00 Beijing Time on August 23, allowing developers to run batch processing and AI agents at reduced rates\. The adjustment aims to optimize the company's compute capacity during traditionally idle periods\. The change follows an August 16 update that raised peak rates fivefold and off\-peak rates 2\.5 times for the V4 Flash model\. Daily token usage for that model had previously surged from 3 trillion to 18 trillion tokens on OpenCode within two weeks of its August 1 debut before the price increase stalled growth\.

## Sources

- [Story source](<https://huggingnews.com/ai/update-deepseek-slashes-weekend-api-costs-to-off-peak-rates-from-aug-23-839a0c76>)
- [Story source](<https://x.com/MegaBlackLabel/status/2091154621154869248>)
- [Supporting source](<https://x.com/MaxForAI/status/2091150619533009361>)
- [Supporting source](<https://x.com/hqmank/status/2091160923457352127>)