SEO content at scale cost calculator

Programmatic SEO generation takes a short structured brief — target keyword, entity data, maybe 300–600 tokens — and produces a full 800–1,200 token article, resulting in the 30/70 input/output ratio. Context stays short at 4k because briefs are self-contained and there is no long retrieved document to hold in window.

Batch scheduling is assumed throughout, so best-effort latency applies and per-token throughput cost dominates economics. cachedPromptPercent is low at 5 because the content brief changes every call. The main cost lever is output length: tighter word-count instructions directly reduce spend. Instruction-following quality matters more than raw benchmark scores — models that respect length constraints and heading structure reduce post-processing cleanup. Fast 70B models are the sweet spot; 405B adds marginal quality for 5× the cost.

Recommended models