DeepSeek V4.1-Flash-1017 tops OpenRouter weekly usage, up 801 percent
OpenRouter data published on Oct 26 shows global large model usage of 188 trillion tokens for the week from Oct 19 to 25, up 14.63 percent on the prior week. Chinese models accounted for 93.54 trillion tokens, up 18.14 percent, against 22.47 trillion for US models, up 29.44 percent, keeping Chinese models ahead of US models for 26 consecutive weeks. DeepSeek's V4.1-Flash-1017, updated on Oct 17, took the top model position with 33.6 trillion tokens, up 801 percent, even as the previous generation V4.1 Flash from Sep 10 fell 45 percent, showing a rapid migration to the newer version. All top five positions went to Chinese models, with MiniMax's M3.1 Flash released on the evening of Oct 19 third at 13.7 trillion tokens, Zhipu's GLM 5.3 Flash fourth at 13.2 trillion up 8 percent, and Tencent's Hy4 preview fifth at 8.97 trillion up 19 percent. Among overseas models, OpenAI's GPT-6 Luna ranked sixth at 8.35 trillion tokens and Anthropic's Claude Opus 5.5 seventh at 8.27 trillion. OpenRouter notes the ranking counts prompt and completion tokens including reasoning tokens, and covers only traffic routed through the platform rather than vendor-owned APIs, cloud platforms, private deployments or consumer applications.