DeepSeek's weekly OpenRouter usage topped OpenAI, Google, Anthropic and xAI combined
aigclink · x · 2026-10-08
According to OpenRouter data (cited by socialcapital), DeepSeek models processed 36.8 trillion tokens in the last week of September — more than OpenAI, Google, Anthropic and xAI combined, averaging 5.3T tokens per day.
Chinese models: DeepSeek 36.8T, Zhipu 14T, Xiaomi 11.9T, Kimi 1.8T, MiniMax 1.3T. US models: OpenAI 19.5T, Google 6.8T, Anthropic 6.1T, xAI 0.6T. Chinese models totaled 65.8T tokens, roughly 2x the US total of 33T.
More from Infra
- Enterprise AI trends toward model routers, not one giant model — ingliguori · 2026-10-08
- Crowdsourced harness x model benchmark: 3090 beats 5090 with qwen3.8-flash-next config — dh7net · 2026-10-08
- llama.cpp distributes inference across heterogeneous devices: MiMo 2.6 Flash at 40 tok/s over 10 GbE — joao_gante · 2026-10-08
- Java-based jitLLM claims 90% of llama.cpp perf on NVIDIA GPUs via TornadoVM CUDA compilation — mikebmx1 · 2026-10-08
- Taiwan OSATs Ordered $11B of Equipment in 9 Months, Topping 2019–2025 Combined — zephyr_z9 · 2026-10-08
- NVIDIA Announces GTC Berlin: Jensen Huang Keynote Set for October 21 — GavinSBaker · 2026-10-08