Nebius cached-token pricing sends its OpenRouter traffic vertical, from under 20B to 67B tokens a day
AccBalanced · x · 2026-10-25
Nebius Token Factory's new cached-token pricing made its OpenRouter traffic go vertical: mostly under 20B tokens/day since July, 67B yesterday, pacing 58B on a Saturday.
- Pricing gap: cached reads on DeepSeek V4.1 Flash 1017 cost $0.006 per 1M tokens vs $0.30 for regular input — repeated context (system prompts, agent history) now bills at a fraction.
- The underrated mechanism is sticky routing: when a provider's cache reads are cheaper, OpenRouter routes the rest of that session back to it. Without cache pricing Nebius couldn't hold an agent session; now it can.
- Speed was already there (738 tok/s on the same model); price and stickiness complete the package.
Charts show DeepSeek V4.1 Flash 1017 leading daily usage at 7.98B tokens, Qwen3 Embedding 8B at 7.86B, total 21.7B with a daily pace of 57.9B (+36.2B).
Related event: Nebius cached-token pricing boosts OpenRouter daily traffic to 67B tokens(2 posts)→
More from Infra
- Nebius hits 67B tokens/day on OpenRouter, but rivals process up to 2T daily — AccBalanced · 2026-10-25
- OpenRouter analysis finds evidence of spoofing-style price manipulation by inference providers — AccBalanced · 2026-10-25
- EmbeddedLLM receives AMD MI350X server donation to advance vLLM ROCm work — AccBalanced · 2026-10-25
- 106 Tesla Megapacks to replace a former Belgian coal plant with 400 MWh storage — XFreeze · 2026-10-25
- Best setup to run Qwen3.8 27B locally on an M4 Pro 48GB Mac? — 9kSs · 2026-10-25
- Security researchers find 32 NVIDIA B300 GPUs on Chinese networks in scan of 12,000+ exposed GPUs — AccBalanced · 2026-10-25