Preserve coding-plan provenance in qualification

docs/AI.md

@@ -415,6 +415,12 @@ is not represented as a zero-dollar API price. The report keeps its monetary
 cost unknown and marks the per-run metered-API budget as not applicable; plan
 limits, provider errors, latency, and any usage the CLI exposes remain visible.
 
+Z.ai requests sent through a configured `/api/coding/paas/` endpoint are
+recorded as `inference_route=coding_plan_allowance`. Qualification keeps their
+per-run monetary cost unknown and the metered-API dollar budget not applicable,
+while still scoring tokens, latency, provider or plan failures, and model
+quality. The standard Z.ai `/api/paas/` endpoint remains a `metered_api` route.
+
 ### Models
 
 Pulse uses model identifiers in the form: `provider:model-name`

docs/release-control/v6/internal/subsystems/ai-runtime.md

@@ -297,6 +297,9 @@ pricing on `metered_api` remains a fail-closed dollar-budget failure. A
 monetary cost as unknown with `budget_applicable=false`; it must not invent a
 zero-dollar price, and latency, provider/plan errors, tool efficiency, and any
 usage counts exposed by the transport remain scored and reportable.
+The configured Z.ai `/api/coding/paas/` endpoint is likewise reported as
+`coding_plan_allowance`, with unknown monetary cost and no per-run metered-API
+dollar budget; the standard `/api/paas/` endpoint remains `metered_api`.
 
 ## Canonical Files
 

internal/ai/qualification/client.go

@@ -54,6 +54,7 @@ type AISettings struct {
 	PatrolEnabled             bool   `json:"patrol_enabled"`
 	CodexSubscriptionEnabled  bool   `json:"codex_subscription_enabled"`
 	ClaudeSubscriptionEnabled bool   `json:"claude_subscription_enabled"`
+	ZaiBaseURL                string `json:"zai_base_url"`
 }
 
 type PulseVersion struct {

internal/ai/qualification/report.go

@@ -6,6 +6,7 @@ import (
 	"encoding/json"
 	"fmt"
 	"html"
+	"net/url"
 	"os"
 	"path/filepath"
 	"regexp"
@@ -52,6 +53,18 @@ func inferenceRouteForProvider(provider string) string {
 	}
 }
 
+func inferenceRouteForProviderEndpoint(provider, baseURL string) string {
+	provider = strings.ToLower(strings.TrimSpace(provider))
+	if provider != "zai" {
+		return inferenceRouteForProvider(provider)
+	}
+	parsed, err := url.Parse(strings.TrimSpace(baseURL))
+	if err == nil && strings.Contains(strings.ToLower(parsed.Path), "/coding/paas/") {
+		return "coding_plan_allowance"
+	}
+	return inferenceRouteForProvider(provider)
+}
+
 type RunReport struct {
 	SchemaVersion         string                                      `json:"schema_version"`
 	RunID                 string                                      `json:"run_id"`

internal/ai/qualification/report_inference_route_test.go

@@ -0,0 +1,25 @@
+package qualification
+
+import "testing"
+
+func TestInferenceRouteForProviderEndpoint(t *testing.T) {
+	tests := []struct {
+		name     string
+		provider string
+		baseURL  string
+		want     string
+	}{
+		{name: "Zai coding plan", provider: "zai", baseURL: "https://api.z.ai/api/coding/paas/v4", want: "coding_plan_allowance"},
+		{name: "Zai coding completions", provider: "ZAI", baseURL: "https://api.z.ai/api/coding/paas/v4/chat/completions", want: "coding_plan_allowance"},
+		{name: "Zai metered API", provider: "zai", baseURL: "https://api.z.ai/api/paas/v4", want: "metered_api"},
+		{name: "unparseable Zai endpoint", provider: "zai", baseURL: "://", want: "metered_api"},
+		{name: "local subscription agent", provider: "codex-subscription", baseURL: "", want: "local_subscription_agent"},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			if got := inferenceRouteForProviderEndpoint(tt.provider, tt.baseURL); got != tt.want {
+				t.Fatalf("inferenceRouteForProviderEndpoint(%q, %q) = %q, want %q", tt.provider, tt.baseURL, got, tt.want)
+			}
+		})
+	}
+}

internal/ai/qualification/runner.go

@@ -175,7 +175,7 @@ func (r *QualificationRunner) Run(ctx context.Context) (report RunReport, termin
 			PulseVersion: version.Version,
 			PulseBaseURL: r.config.Client.config.BaseURL,
 			DockerTarget: dockerTargetLabel(r.config.Lab.target),
-			Model:        model, Provider: provider, InferenceRoute: inferenceRouteForProvider(provider), ChallengeNonce: r.config.ChallengeNonce,
+			Model:        model, Provider: provider, InferenceRoute: inferenceRouteForProviderEndpoint(provider, settings.ZaiBaseURL), ChallengeNonce: r.config.ChallengeNonce,
 			CapturedAt: time.Now().UTC(),
 		}
 		return nil

internal/ai/qualification/scorer_test.go

@@ -61,6 +61,16 @@ func TestScoreRunAppliesDollarBudgetOnlyToMeteredAPIRoutes(t *testing.T) {
 	if strings.Contains(strings.Join(subscription.GateFailures, "\n"), "cost budget") {
 		t.Fatalf("subscription allowance was treated as metered API spend: %+v", subscription.GateFailures)
 	}
+	codingPlan := ScoreRun(ScoringInput{
+		Manifest: manifest, Provider: "zai", Model: "zai:glm-5.2",
+		InferenceRoute: "coding_plan_allowance",
+	})
+	if codingPlan.Cost.Known || codingPlan.Cost.BudgetApplicable || codingPlan.Cost.BillingBasis != "coding_plan_allowance" {
+		t.Fatalf("unexpected coding-plan cost semantics: %+v", codingPlan.Cost)
+	}
+	if strings.Contains(strings.Join(codingPlan.GateFailures, "\n"), "cost budget") {
+		t.Fatalf("coding-plan allowance was treated as metered API spend: %+v", codingPlan.GateFailures)
+	}
 
 	metered := ScoreRun(ScoringInput{
 		Manifest: manifest, Provider: "unpriced-api", Model: "unpriced-api:model",

scripts/release_control/ai_runtime_docs_policy_test.py

@@ -151,6 +151,9 @@ class AIRuntimeDocsPolicyTest(unittest.TestCase):
         self.assertIn("Pulse retains tool execution and policy enforcement", normalized_content)
         self.assertIn("not represented as a zero-dollar API price", normalized_content)
         self.assertIn("per-run metered-API budget as not applicable", normalized_content)
+        self.assertIn("inference_route=coding_plan_allowance", content)
+        self.assertIn("per-run monetary cost unknown", normalized_content)
+        self.assertIn("standard Z.ai `/api/paas/` endpoint remains a `metered_api` route", normalized_content)
 
     def test_public_ai_privacy_copy_discloses_outbound_usage_telemetry(self) -> None:
         content = read_repo_text("docs/AI.md")