Chinese AI lab DeepSeek released a new coding model, V4 Flash, on Friday, charging about 28 cents for the same amount of output that costs $25 on Anthropic's Claude Opus 4.8, a 99% discount. V4 Flash performs close to Opus 4.8 on tests of complex coding and autonomous software tasks and debuted ahead of it on Arena.ai's crowdsourced leaderboard for front-end coding, while delivering the best performance for its price among any model in its class. DeepSeek previously ignited a market meltdown last January by building a world-class AI model with fewer resources than U.S. rivals. July brought a full-scale price war as Chinese models like Kimi K3 entered the U.S. market. OpenAI cut the price of GPT-5.6 Luna by 80% on Thursday, three weeks after its launch. Google released three new Gemini "flash" models focused on efficiency. SpaceXAI released Grok 4.5 at the same price OpenAI originally charged for Luna before the cut. Meta released Muse Spark 1.1, a closed-source model priced aggressively. Anthropic remains a holdout, keeping premium pricing. As performance gaps shrink, buyers gain leverage to shop on price, a phenomenon called "diminishing model returns." This could create a market for "intelligent routers" that automatically choose the best model per task.