From 16:00 UTC on 22 August, off-peak rates cover all of Saturday and Sunday. Off-peak V4-Pro output costs $1.98 per million tokens; before 16 August the flat rate was $0.87.
Aug 23, 2026
V4-Flash-Vision-Exp is experimental, API-only, and scored entirely on agentic and coding evaluations.
Aug 21, 2026
Release candidate 8 wires up an image transport path — for third-party vision models. Chinese coverage read it as a V4 spoiler.
Aug 20, 2026
DeepSeek twice warned of a "significant" increase without naming a number or a date. Both have now arrived — and the largest rise is not the one being quoted.
Aug 17, 2026
DeepSeek V4 Pro and Zhipu's GLM-5.3 sit in a 'frontier models' tier inside an Alibaba product. GLM-5.3 shipped Saturday morning in Beijing and was selectable by Sunday lunchtime.
Aug 16, 2026
Two days after DeepSeek released Harness as a developer preview, China's largest chat platform made it a plugin — npm install, launch, scan a QR code.
Aug 16, 2026
V4-Pro-0813 adds the Responses API and tool calls. The more consequential line is a note on the pricing page about an imminent increase.
Aug 13, 2026
Shanghai's flagship humanoid listing raised 6.1bn yuan against a 4.2bn target, with China's most visible AI lab on the strategic placement list.
Aug 9, 2026
The company that started the AI price war told developers on 6 August that its API is about to get substantially more expensive. It published no rate card, no effective date and no explanation.
Aug 6, 2026
V4-Flash went official on Friday at roughly a third of V4-Pro's input price, and scores above it across DeepSeek's own agent benchmarks.
Jul 31, 2026
Unit 42 traced a Chinese-speaking operator running intrusions through a Telegram-orchestrated agent built on open Chinese model weights. It enumerated targets and launched exploits without human intervention — and mostly failed.
Jul 30, 2026
The lab verbally told some would-be backers it will not sign in the coming days, shelving a raise of at least 10 billion yuan at a pre-money valuation of at least 480 billion — after leaked remarks attributed to its founder went viral.
Jul 26, 2026
The deepseek-chat and deepseek-reasoner aliases stopped working at 15:59 UTC on Friday, forcing every remaining integration onto the V4 line after a three-month notice.
Jul 25, 2026
The AI Security Institute found open-weight systems like GLM-5.2 and DeepSeek V4-Pro are closing fast on the best closed models in offensive-security tasks, shrinking the window defenders have to prepare.
Jul 18, 2026
The Hangzhou lab behind the open-weight model that rattled the industry is reportedly seeking fresh capital at a roughly $71 billion valuation — barely a month after its first outside round — while preparing a mainland-China listing.
Jul 15, 2026
The Chinese lab that shook the AI world with cheap models is now going after the silicon — a SMIC-fabricated inference chip meant to cut its dependence on both Nvidia and Huawei.
Jul 8, 2026
A year ago Chinese models carried under 2% of OpenRouter traffic. Now four of the top five models are Chinese, Xiaomi's MiMo-V2-Pro processes more than 4 trillion tokens a week — and price is the weapon.
Jul 7, 2026
