Add cost chart to card

README.md

@@ -106,6 +106,8 @@ Held-out eval of 1,977 conversations, zero train/eval leakage, macro-averaged (e
 
 ![ZeroProof e-commerce intent evaluation: accuracy, per-action accuracy, and cost to serve](1b-v1.png)
 
+![Accuracy versus cost to serve: zeroproof-ecommerce-1b near frontier accuracy at about one hundredth the serving cost](cost-chart.png)
+
 - Fine-tuning takes intent-type accuracy from **12 to 77.5** (macro), landing at **96% of GPT-5's 80.2**.
 - On structured detail extraction it **exceeds the full frontier panel**: 64 against Sonnet 5's 56, Opus 4.8's 53, and GPT-5's 41.
 - On the no-action boundary it errs toward inaction: **9.2%** of true no-action conversations are marked actionable, with remaining errors concentrated in the safe direction (missed actions, not invented ones).