The Math: Claiming 100T Tokens/Day Would Need ~580K GPUs
teortaxesTex · x · 2026-08-21
The author runs the unit-economics math on claims of massive daily token throughput. In serious agentic sessions, input+output tokens are under 5% of total processing — he observes 1% on the DeepSeek platform at 0.5-1B tokens/day, so 1T "honest tokens" is just one small cluster.
He further estimates that running a 5.3-scale model at DS-V4 rate (2K tokens/GPU/s at 80 tps) would require 580K GPUs to hit 100T tokens/day — implying the claimed token count likely isn't output tokens, or the number is questionable.
More from Infra
- Muse Spark Open runs locally on a Mac M3 Ultra via llama.cpp: frontier model on desktop — pcuenq · 2026-10-16
- Muse-Spark open-weights drop: 816B-A42B MoE model gets 359GB MLX 3-bit build — ns123abc · 2026-10-16
- vLLM ships Day-0 LoRA RL for Muse Spark Open, lifting AIME from 0.56 to 0.85 — vllm_project · 2026-10-16
- DSpark draft model for Muse Spark Open proposes 7 tokens per pass with 23-42% acceptance — vllm_project · 2026-10-16
- Cloudflare Stream's WebRTC ultra-low-latency live streaming is now Generally Available — ritakozlov · 2026-10-16
- YC-backed Overshoot launches inference optimization, claims $5M saved for AI teams — ycombinator · 2026-10-16