DeepSeek Flash Traffic Halves After Price Hike, Leaving Nearly 10 Trillion Token Daily Gap

OpenCode CEO Jay V stated that since DeepSeek V4 Flash launched on August 1, daily token consumption on OpenCode surged from approximately 3 trillion to 18 trillion, a 6-fold increase within two weeks, approaching the daily processing volume of the entire OpenRouter platform. Its pricing is roughly 1/350 of Fable, 1/175 of GPT-5.6 Sol, and 1/70 of Claude Sonnet, leading users to view it as "AI so cheap it doesn't need metering." After DeepSeek raised peak-hour prices by 5x (and 2.5x off-peak) on August 16, the model's daily token consumption fell by over 50% from its peak. Jay V believes the price hike was a reactive move to address rising GPU costs. OpenCode is currently seeking suppliers who can handle the traffic at original prices, requiring approximately 1,000 B300 units to support current demand. DeepSeek Flash has exposed the "impossible trinity" of the AI model market—cheap, powerful, and stably scalable supply are hard to achieve simultaneously. Following the price hike, OpenCode Go has reduced quotas from $60 to $15, then restored them to $30, and introduced peak-off-peak rate mechanisms.