Provider cost + 6%, itemized. Answer a question strictly from the context you provide — grounded, no hallucinated sources. ≈Opus-4.8 accuracy at ~1–5% of its cost (benchmark: relaystation.ai/llm-validation). 1¢ x402 min; remainder auto-credits — relaystation.ai/penny