Newsletter Subscribe
Enter your email address below and subscribe to our newsletter
[forminator_form id="25163"]

bloombergqz+1bloombergThe AI pricing landscape shifted in opposite directions this month as DeepSeek announced steep price increases for its V4 model line and OpenAI made its GPT-5.6 Luna model the free default for ChatGPT users — moves that underscore the intensifying competition between Chinese and American AI providers over cost, access, and market share.
DeepSeek announced on August 13 that it would raise API prices for its V4-Flash and V4-Pro models effective August 16, introducing a new peak and off-peak billing structure the company said would "allocate resources more reasonably." For V4-Flash, output tokens rose from $0.28 per million to $1.32 during peak hours — a roughly 370% increase — and $0.66 during off-peak hours. V4-Pro output tokens climbed to $3.96 per million at peak, more than four times the previous flat rate of $0.87. Peak hours are defined as 01:00–04:00 and 06:00–10:00 UTC, with off-peak rates set at half the peak price.qz+2
The increases range from 50% to more than 1,100% depending on the model, token type, and timing. Even at the new rates, DeepSeek remains far cheaper than Anthropic's Claude Opus 5 at $25 per million output tokens or OpenAI's GPT-5.6 Sol at $30, but the move narrows what had been a defining cost advantage for the Chinese lab.qz
Days earlier, OpenAI announced on August 5 that GPT-5.6 Luna would become the default model for Free and Go plan users during the week of August 6, replacing GPT-5.5 Instant. The company also said unlimited text chats and a new "Think" button for harder questions would follow the next week, subject to abuse guardrails. File uploads, images, and other tools remain subject to separate limits.flowith+2
Luna sits at the bottom of OpenAI's GPT-5.6 family, priced at $1 input and $6 output per million tokens on the API. Making it the free consumer default represents OpenAI's bid to retain everyday users as competitors slash prices and expand access.appwrite
The diverging strategies reflect a broader tension in the AI industry. According to Bloomberg, Chinese models overtook US platforms globally in June on OpenRouter, accounting for more than 60% of token share last month, driven largely by aggressive pricing from labs including DeepSeek and Moonshot. DeepSeek's price increases suggest the company may be pivoting from pure land-grab pricing toward sustainability, even as Bloomberg Intelligence analyst Robert Lea noted that none of China's AI labs have a clear path to profit under their current models.bloomberg
For developers who built on DeepSeek's rock-bottom rates, the adjustment forces a recalculation. As one widely cited comparison noted, a coding agent processing 10 million output tokens monthly on V4-Pro now costs roughly $40 at peak hours rather than the previous $9 — still a fraction of frontier US pricing, but no longer an order-of-magnitude gap.tech-insider