feat: route all AI inference through OpenRouter GLM 5.3 Flash

docs/architecture.md

@@ -47,7 +47,7 @@ Client-side environment selection is a developer convenience, not authorization.
 2. The client reads the current Supabase session.
 3. It invokes `ai` or `ai-stream` with selected controls and active source context.
 4. The Edge Function validates authentication, input bounds, supported model and control combinations, access limits, and feature-specific rules.
-5. The backend optionally gathers research, builds prompts, calls a provider, validates output, and records usage.
+5. The backend optionally gathers research, builds prompts, calls an allowlisted model through OpenRouter, validates output, and records the provider-reported usage and charge.
 6. The client validates or transforms the response and updates the relevant store.
 
 [`supabase/functions/ai/index.ts`](../supabase/functions/ai/index.ts) serves non-streaming structured tools and practice modes. [`supabase/functions/ai-stream/index.ts`](../supabase/functions/ai-stream/index.ts) serves Ask over server-sent events.
@@ -83,7 +83,9 @@ Supabase Auth owns user sessions. Protected routes provide a client navigation b
 
 Checkout and portal functions create Stripe sessions after authenticated ownership and redirect validation. The webhook updates subscription state and records event processing for idempotency. AI handlers record provider usage and calculated cost using shared pricing logic.
 
-Provider pricing evidence is retained under [`docs/pricing-snapshots/`](pricing-snapshots/). The snapshots are evidence, not runtime configuration.
+All model requests use the single OpenRouter backend route and `OPENROUTER_API_KEY`. The only selectable and default model is `z-ai/glm-5.3-flash`; the shared allowlist and request controls live in [`supabase/functions/_shared/model-catalog.ts`](../supabase/functions/_shared/model-catalog.ts). OpenRouter usage remains raw in the audit row, and the reported charge is unknown when the provider does not return a complete cost.
+
+Historical provider pricing evidence is retained under [`docs/pricing-snapshots/`](pricing-snapshots/). Those snapshots support legacy calculator records and are evidence, not runtime configuration for the OpenRouter route.
 
 ## Observability
 

package.json

@@ -18,7 +18,7 @@
     "preview": "vite preview",
     "test": "vitest run",
     "test:watch": "vitest",
-    "test:edge": "deno test --allow-read --allow-env=DEMO_CODE_PEPPER,EXA_API_KEY,SUPABASE_URL,SUPABASE_SERVICE_ROLE_KEY,SPARRING_INTEGRITY_SECRET,OPENAI_API_KEY,GOOGLE_API_KEY,ANTHROPIC_API_KEY --allow-net=127.0.0.1 supabase/functions/",
+    "test:edge": "deno test --allow-read --allow-env=DEMO_CODE_PEPPER,EXA_API_KEY,SUPABASE_URL,SUPABASE_SERVICE_ROLE_KEY,SPARRING_INTEGRITY_SECRET,OPENROUTER_API_KEY --allow-net=127.0.0.1 supabase/functions/",
     "test:db": "supabase test db",
     "test:e2e": "playwright test --project=desktop-chromium --project=mobile-chromium",
     "test:e2e:webkit": "playwright test --project=webkit",

src/components/ModelSelector.tsx

@@ -1,7 +1,8 @@
 import { Cpu } from 'lucide-react';
 import { Select, SelectContent, SelectItem, SelectTrigger, SelectValue } from '@/components/ui/select';
 import { Badge } from '@/components/ui/badge';
-import { DEFAULT_PROVIDERS } from '@/lib/ai/providers/types';
+import { DEFAULT_MODEL, DEFAULT_PROVIDERS, isReasoningEffort } from '@/lib/ai/providers/types';
+import type { ReasoningEffort } from '@/lib/ai/providers/types';
 import { useSettingsStore, type FeatureTab } from '@/lib/stores/settings-store';
 import { useSubscription } from '@/hooks/use-subscription';
 import { Label } from '@/components/ui/label';
@@ -15,20 +16,29 @@ interface ModelSelectorProps {
 }
 
 const allModels = DEFAULT_PROVIDERS
-  .filter(p => p.id === 'google' || p.id === 'openai' || p.id === 'anthropic')
   .flatMap(p => p.models.map(m => ({ ...m, provider: p.id, providerName: p.name })));
 
+const reasoningOptions: { value: ReasoningEffort; labelKey: string; fallback: string }[] = [
+  { value: 'low', labelKey: 'modelSelector.reasoningLow', fallback: 'Low' },
+  { value: 'medium', labelKey: 'modelSelector.reasoningMedium', fallback: 'Medium' },
+  { value: 'high', labelKey: 'modelSelector.reasoningHigh', fallback: 'High' },
+];
+
 export function ModelSelector({ feature, className, showLabel = false }: ModelSelectorProps) {
   const { t } = useTranslation();
   const { featureConfigs, setFeatureConfig } = useSettingsStore();
-  const { isPro } = useSubscription();
+  const { isPro, loading: subscriptionLoading } = useSubscription();
   const config = featureConfigs?.[feature];
-  const currentModel = config?.model || 'gemini-3-flash-preview';
+  const selectedModel = allModels.find((model) => model.id === config?.model);
+  const currentModel = selectedModel?.id ?? DEFAULT_MODEL;
+  const currentReasoning = (isPro || subscriptionLoading) && config?.reasoningEffort && isReasoningEffort(config.reasoningEffort)
+    ? config.reasoningEffort
+    : 'low';
 
   return (
     <div className={cn("space-y-2", className)}>
       {showLabel && (
-        <Label className="flex items-center gap-2 text-sm font-medium">
+        <Label className="flex h-5 items-center gap-2 text-sm font-medium">
           <Cpu className="h-4 w-4 text-muted-foreground" />
           {t('modelSelector.label', 'AI model')}
         </Label>
@@ -62,6 +72,33 @@ export function ModelSelector({ feature, className, showLabel = false }: ModelSe
           ))}
         </SelectContent>
       </Select>
+      <div className="space-y-2">
+        <Label className="text-sm font-medium">
+          {t('modelSelector.reasoning', 'Reasoning')}
+        </Label>
+        <Select
+          value={currentReasoning}
+          disabled={!isPro}
+          onValueChange={(value) => {
+            if (!isPro || !isReasoningEffort(value)) return;
+            setFeatureConfig(feature, { reasoningEffort: value });
+          }}
+        >
+          <SelectTrigger
+            className="h-10 w-full text-sm"
+            aria-label={t('modelSelector.reasoning', 'Reasoning')}
+          >
+            <SelectValue />
+          </SelectTrigger>
+          <SelectContent>
+            {reasoningOptions.map((option) => (
+              <SelectItem key={option.value} value={option.value} className="text-xs">
+                {t(option.labelKey, option.fallback)}
+              </SelectItem>
+            ))}
+          </SelectContent>
+        </Select>
+      </div>
     </div>
   );
 }

src/lib/ai/client.ts

@@ -369,7 +369,7 @@ async function callSupabaseAI<T = unknown>(
     ? boundCrossfireRequestSources(selectedSources)
     : selectedSources;
 
-  const provider = config.provider === 'lovable' ? 'openai' : config.provider;
+  const provider = config.provider === 'mock' ? 'mock' : 'openrouter';
   captureProductEvent('ai_request_started', {
     feature,
     provider,
@@ -420,6 +420,7 @@ async function callSupabaseAI<T = unknown>(
       provider,
       model: config.model,
       temperature: config.temperature,
+      reasoningEffort: config.reasoningEffort ?? 'low',
       customSystemPrompt: options.includeCustomSystemPrompt === false ? null : settings.customSystemPrompt || null,
       sources: activeSources,
     };
@@ -1236,7 +1237,7 @@ async function sparringRequestKey(
 ): Promise<string> {
   const { data } = await supabase.auth.getSession();
   const authScope = data.session?.user?.id ?? data.session?.access_token ?? 'signed-out';
-  return JSON.stringify([authScope, input, config.provider, config.model, config.temperature]);
+  return JSON.stringify([authScope, input, config.provider, config.model, config.temperature, config.reasoningEffort]);
 }
 
 /** Generate the opponent's next speech for a validated sparring turn. */

src/lib/ai/providers/types.ts

@@ -1,62 +1,39 @@
 // src/lib/ai/providers/types.ts
 
+import { MODEL_CATALOG } from '../../../../supabase/functions/_shared/model-catalog';
+import type { ReasoningEffort } from '../../../../supabase/functions/_shared/model-catalog';
+
+export { DEFAULT_MODEL, DEFAULT_PRO_MODEL, isReasoningEffort } from '../../../../supabase/functions/_shared/model-catalog';
+export type { ReasoningEffort } from '../../../../supabase/functions/_shared/model-catalog';
+
 export interface AIProvider {
   id: string;
   name: string;
-  models: AIModel[];
+  models: readonly AIModel[];
   isAvailable: () => boolean;
-  requiresKey?: boolean;
 }
 
 export interface AIModel {
   id: string;
   name: string;
-  description: string;
   maxTokens: number;
-  isProOnly?: boolean;
+  isProOnly: boolean;
+  supportsTemperature: boolean;
+  supportedEfforts: readonly ReasoningEffort[];
 }
 
 export interface ProviderConfig {
   provider: string;
   model: string;
   temperature: number;
-  apiKey?: string;
-  baseUrl?: string;
+  reasoningEffort?: ReasoningEffort;
 }
 
 export const DEFAULT_PROVIDERS: AIProvider[] = [
   {
-    id: 'openai',
-    name: 'OpenAI',
-    models: [
-      { id: 'gpt-5.5', name: 'GPT-5.5', description: '', maxTokens: 128000, isProOnly: true },
-      { id: 'gpt-5.4', name: 'GPT-5.4', description: '', maxTokens: 128000, isProOnly: true },
-      { id: 'gpt-5.4-mini', name: 'GPT-5.4 Mini', description: '', maxTokens: 128000 },
-    ],
-    isAvailable: () => true,
-    requiresKey: true,
-  },
-  {
-    id: 'anthropic',
-    name: 'Anthropic',
-    models: [
-      { id: 'claude-opus-4-7', name: 'Claude Opus 4.7', description: '', maxTokens: 200000, isProOnly: true },
-      { id: 'claude-sonnet-4-6', name: 'Claude Sonnet 4.6', description: '', maxTokens: 200000, isProOnly: true },
-      { id: 'claude-haiku-4-5-20251001', name: 'Claude Haiku 4.5', description: '', maxTokens: 200000 },
-    ],
-    isAvailable: () => true,
-    requiresKey: true,
-  },
-  {
-    id: 'google',
-    name: 'Google Gemini',
-    models: [
-      { id: 'gemini-3.1-pro-preview', name: 'Gemini 3.1 Pro', description: '', maxTokens: 2000000, isProOnly: true },
-      { id: 'gemini-3.5-flash', name: 'Gemini 3.5 Flash', description: '', maxTokens: 1000000 },
-      { id: 'gemini-3-flash-preview', name: 'Gemini 3 Flash', description: '', maxTokens: 1000000 },
-      { id: 'gemini-3.1-flash-lite-preview', name: 'Gemini 3.1 Flash Lite', description: '', maxTokens: 1000000 },
-    ],
+    id: 'openrouter',
+    name: 'OpenRouter',
+    models: MODEL_CATALOG,
     isAvailable: () => true,
-    requiresKey: true,
   },
 ];

src/lib/ai/streaming.ts

@@ -24,7 +24,6 @@ export type ParsedSSEEvent = {
 };
 
 export function isAskStreamingSupported(config: ProviderConfig, options: AskStreamOptions = {}) {
-  const provider = config.provider === 'lovable' ? 'openai' : config.provider;
   const advancedStructuredMode = Boolean(
     (options.mode ?? 'general') !== 'general' ||
     options.citationMode ||
@@ -33,7 +32,7 @@ export function isAskStreamingSupported(config: ProviderConfig, options: AskStre
     options.argumentMapMode
   );
 
-  return !advancedStructuredMode && ['openai', 'google', 'anthropic'].includes(provider);
+  return !advancedStructuredMode && config.provider === 'openrouter';
 }
 
 function parseEventBlock(block: string): ParsedSSEEvent | null {
@@ -115,7 +114,7 @@ export async function streamAskChat(
   options: AskStreamOptions,
   handlers: AskStreamHandlers
 ): Promise<AskOutput> {
-  const provider = config.provider === 'lovable' ? 'openai' : config.provider;
+  const provider = config.provider === 'mock' ? 'mock' : 'openrouter';
   const requestId = `stream_${Date.now()}_${Math.random().toString(36).slice(2, 8)}`;
   const startTime = performance.now();
   captureProductEvent('ai_request_started', {
@@ -203,6 +202,7 @@ export async function streamAskChat(
         provider,
         model: config.model,
         temperature: config.temperature,
+        reasoningEffort: config.reasoningEffort ?? 'low',
       }),
       signal: controller.signal,
     });