LLM inference cost — and missing latency data under long context

Pain point

Cost of inference is a live concern for SaaS margins; the top reply flags the real gap: **"The number