{"agent_only_efficiency": {"baseline": {"tokens_per_resolved": null, "usd_per_resolved": null}, "marginal": {"tokens_per_resolved": null, "usd_per_resolved": null}}, "baseline": {"effective_latency_ms": 593106, "effective_tokens": 1098747, "effective_usd": 0.0, "false_stop_rate": null, "false_stops": 0, "governance_latency_ms": 0, "governance_tokens": 0, "governance_usd": 0.0, "latency_ms": 593106, "repeated_calls": 0, "resolve_rate": 0.0, "resolved": 0, "reviewed_stops": 0, "tasks": 3, "tokens": 1098747, "tool_calls": 33, "usd": 0.0}, "benchmark": "public-agent-benchmark-comparison-v3", "efficiency": {"baseline": {"tokens_per_resolved": null, "usd_per_resolved": null}, "marginal": {"tokens_per_resolved": null, "usd_per_resolved": null}}, "governance": {"latency_ms": 7058, "tokens": 0, "usd": 0.0}, "gross_savings": {"confidence_level": 0.95, "latency_percent": 5.8, "repeated_calls_percent": 0.0, "tokens_confidence_interval": [21.52, 28.07], "tokens_percent": 24.93, "tool_calls_percent": 3.03, "usd_percent": 0.0}, "intervention": {"eligible_for_support": false, "graceful_irrelevance": true, "minimum_net_token_savings_percent": 0.0, "net_positive": true, "status": "pass_through"}, "marginal": {"effective_latency_ms": 565768, "effective_tokens": 824839, "effective_usd": 0.0, "false_stop_rate": null, "false_stops": 0, "governance_latency_ms": 7058, "governance_tokens": 0, "governance_usd": 0.0, "latency_ms": 558710, "repeated_calls": 0, "resolve_rate": 0.0, "resolved": 0, "reviewed_stops": 0, "tasks": 3, "tokens": 824839, "tool_calls": 32, "usd": 0.0}, "net_savings": {"confidence_level": 0.95, "latency_percent": 4.61, "repeated_calls_percent": 0.0, "tokens_95pct_ci": [21.52, 28.07], "tokens_confidence_interval": [21.52, 28.07], "tokens_percent": 24.93, "tool_calls_percent": 3.03, "usd_percent": 0.0}, "quality": {"false_stop_rate": null, "false_stops": 0, "false_stops_acceptable": true, "has_verified_success": false, "max_false_stop_rate": 0.0, "non_inferiority_margin_pp": 1.0, "preserved_within_margin": true, "preserved_within_one_pp": true, "recoveries": 0, "regressions": 0, "resolved_delta_pp": 0.0, "reviewed_stops": 0}, "savings": {"confidence_level": 0.95, "latency_percent": 4.61, "repeated_calls_percent": 0.0, "tokens_95pct_ci": [21.52, 28.07], "tokens_confidence_interval": [21.52, 28.07], "tokens_percent": 24.93, "tool_calls_percent": 3.03, "usd_percent": 0.0}, "tasks": 3}
