feat(gui): expose requestPacing.maxConcurrentRequests in provider settings

gui/src/components/provider-workspace/ProviderSettings.tsx

@@ -28,7 +28,7 @@ const ADAPTERS = ["openai-responses", "openai-chat", "anthropic", "google", "azu
 const EMPTY_MODELS: string[] = [];
 
 type ChoicesStatus = "idle" | "loading" | "ready" | "error";
-type PacingRule = { requestsPerMinute?: number; minIntervalMs?: number };
+type PacingRule = { requestsPerMinute?: number; minIntervalMs?: number; maxConcurrentRequests?: number };
 type PacingStatus = { enabled: boolean; queued: number; nextSlotInMs: number; lastStartedAt?: number; lastModelId?: string };
 type CursorHttpVersion = "http2" | "http1.1";
 
@@ -50,11 +50,12 @@ function positiveInteger(value: string): number | undefined {
 function pacingSignature(value: WorkspaceItem["requestPacing"] | undefined): string {
   const models = Object.entries(value?.models ?? {})
     .sort(([left], [right]) => left.localeCompare(right))
-    .map(([model, rule]) => [model, rule.requestsPerMinute ?? null, rule.minIntervalMs ?? null]);
+    .map(([model, rule]) => [model, rule.requestsPerMinute ?? null, rule.minIntervalMs ?? null, rule.maxConcurrentRequests ?? null]);
   return JSON.stringify([
     value?.enabled === true,
     value?.requestsPerMinute ?? null,
     value?.minIntervalMs ?? null,
+    value?.maxConcurrentRequests ?? null,
     models,
   ]);
 }
@@ -96,10 +97,12 @@ export default function ProviderSettings({
   const [pacingEnabled, setPacingEnabled] = useState(item.requestPacing?.enabled === true);
   const [pacingRpm, setPacingRpm] = useState(() => numberDraft(item.requestPacing?.requestsPerMinute));
   const [pacingDelay, setPacingDelay] = useState(() => numberDraft(item.requestPacing?.minIntervalMs));
+  const [pacingConcurrency, setPacingConcurrency] = useState(() => numberDraft(item.requestPacing?.maxConcurrentRequests));
   const [pacingModels, setPacingModels] = useState<Record<string, PacingRule>>(() => ({ ...(item.requestPacing?.models ?? {}) }));
   const [pacingModelId, setPacingModelId] = useState("");
   const [pacingModelRpm, setPacingModelRpm] = useState("");
   const [pacingModelDelay, setPacingModelDelay] = useState("");
+  const [pacingModelConcurrency, setPacingModelConcurrency] = useState("");
   const [pacingStatus, setPacingStatus] = useState<PacingStatus | null>(null);
 
   /* eslint-disable react-hooks/set-state-in-effect -- intentional form reset when saved provider fields change */
@@ -116,6 +119,7 @@ export default function ProviderSettings({
     setPacingEnabled(item.requestPacing?.enabled === true);
     setPacingRpm(numberDraft(item.requestPacing?.requestsPerMinute));
     setPacingDelay(numberDraft(item.requestPacing?.minIntervalMs));
+    setPacingConcurrency(numberDraft(item.requestPacing?.maxConcurrentRequests));
     setPacingModels({ ...(item.requestPacing?.models ?? {}) });
     setMsg(null);
     setModeMsg(null);
@@ -186,8 +190,9 @@ export default function ProviderSettings({
     enabled: pacingEnabled,
     ...(positiveRpm(pacingRpm) !== undefined ? { requestsPerMinute: positiveRpm(pacingRpm) } : {}),
     ...(positiveInteger(pacingDelay) !== undefined ? { minIntervalMs: positiveInteger(pacingDelay) } : {}),
+    ...(positiveInteger(pacingConcurrency) !== undefined ? { maxConcurrentRequests: positiveInteger(pacingConcurrency) } : {}),
     ...(Object.keys(pacingModels).length > 0 ? { models: pacingModels } : {}),
-  }), [pacingDelay, pacingEnabled, pacingModels, pacingRpm]);
+  }), [pacingConcurrency, pacingDelay, pacingEnabled, pacingModels, pacingRpm]);
 
   const dirty = adapter.trim() !== item.adapter
     || baseUrl.trim() !== item.baseUrl
@@ -235,7 +240,10 @@ export default function ProviderSettings({
     setSaving(true);
     setMsg(null);
     try {
-      if (pacingEnabled && !pacingDraft.requestsPerMinute && !pacingDraft.minIntervalMs && !pacingDraft.models) {
+      if (pacingConcurrency.trim() !== "" && positiveInteger(pacingConcurrency) === undefined) {
+        setMsg({ ok: false, text: t("pws.pacingCapInvalid") }); return false;
+      }
+      if (pacingEnabled && !pacingDraft.requestsPerMinute && !pacingDraft.minIntervalMs && !pacingDraft.maxConcurrentRequests && !pacingDraft.models) {
         setMsg({ ok: false, text: t("pws.pacingRuleRequired") }); return false;
       }
       const pacingOnly = pacingDirty && !dirty;
@@ -322,7 +330,9 @@ export default function ProviderSettings({
     setNote(item.note ?? ""); setAllowPrivateNetwork(item.allowPrivateNetwork ?? false); setLiveModels(savedLiveModels);
     setCursorHttpVersion(savedCursorHttpVersion); setMsg(null);
     setPacingEnabled(item.requestPacing?.enabled === true); setPacingRpm(numberDraft(item.requestPacing?.requestsPerMinute));
-    setPacingDelay(numberDraft(item.requestPacing?.minIntervalMs)); setPacingModels({ ...(item.requestPacing?.models ?? {}) });
+    setPacingDelay(numberDraft(item.requestPacing?.minIntervalMs));
+    setPacingConcurrency(numberDraft(item.requestPacing?.maxConcurrentRequests));
+    setPacingModels({ ...(item.requestPacing?.models ?? {}) });
     setEndpointChoice(matchChoiceId(baseUrlChoices, item.baseUrl));
   };
 
@@ -339,9 +349,21 @@ export default function ProviderSettings({
     const modelId = pacingModelId.trim();
     const rpm = positiveRpm(pacingModelRpm);
     const delay = positiveInteger(pacingModelDelay);
-    if (!modelId || (rpm === undefined && delay === undefined)) return;
-    setPacingModels(current => ({ ...current, [modelId]: { ...(rpm !== undefined ? { requestsPerMinute: rpm } : {}), ...(delay !== undefined ? { minIntervalMs: delay } : {}) } }));
-    setPacingModelId(""); setPacingModelRpm(""); setPacingModelDelay("");
+    const cap = positiveInteger(pacingModelConcurrency);
+    if (pacingModelConcurrency.trim() !== "" && cap === undefined) {
+      setMsg({ ok: false, text: t("pws.pacingCapInvalid") });
+      return;
+    }
+    if (!modelId || (rpm === undefined && delay === undefined && cap === undefined)) return;
+    setPacingModels(current => ({
+      ...current,
+      [modelId]: {
+        ...(rpm !== undefined ? { requestsPerMinute: rpm } : {}),
+        ...(delay !== undefined ? { minIntervalMs: delay } : {}),
+        ...(cap !== undefined ? { maxConcurrentRequests: cap } : {}),
+      },
+    }));
+    setPacingModelId(""); setPacingModelRpm(""); setPacingModelDelay(""); setPacingModelConcurrency("");
   };
 
   return (
@@ -500,6 +522,7 @@ export default function ProviderSettings({
         <div className="pwi-pacing-grid">
           <label className="pwi-settings-field"><span className="pwi-settings-label">{t("pws.pacingRpm")}</span><input className="input" type="number" min="0.016667" step="any" value={pacingRpm} onChange={e => setPacingRpm(e.target.value)} placeholder="38" /></label>
           <label className="pwi-settings-field"><span className="pwi-settings-label">{t("pws.pacingDelay")}</span><input className="input" type="number" min="1" step="1" value={pacingDelay} onChange={e => setPacingDelay(e.target.value)} placeholder="1600" /></label>
+          <label className="pwi-settings-field"><span className="pwi-settings-label">{t("pws.pacingConcurrency")}</span><input className="input" type="number" min="1" step="1" value={pacingConcurrency} onChange={e => setPacingConcurrency(e.target.value)} placeholder="8" /></label>
         </div>
         <p className="pwi-settings-hint">{t("pws.pacingSlowerWins")}</p>
         <div className="pwi-pacing-status" aria-live="polite">
@@ -512,9 +535,14 @@ export default function ProviderSettings({
           <label className="pwi-settings-field"><span className="pwi-settings-label">{t("pws.pacingModel")}</span><input className="input" list={`pacing-models-${item.name}`} value={pacingModelId} onChange={e => setPacingModelId(e.target.value)} /><datalist id={`pacing-models-${item.name}`}>{availableModels.map(model => <option key={model} value={model} />)}</datalist></label>
           <label className="pwi-settings-field"><span className="pwi-settings-label">{t("pws.pacingRpm")}</span><input className="input" type="number" min="0.016667" step="any" value={pacingModelRpm} onChange={e => setPacingModelRpm(e.target.value)} /></label>
           <label className="pwi-settings-field"><span className="pwi-settings-label">{t("pws.pacingDelay")}</span><input className="input" type="number" min="1" step="1" value={pacingModelDelay} onChange={e => setPacingModelDelay(e.target.value)} /></label>
+          <label className="pwi-settings-field"><span className="pwi-settings-label">{t("pws.pacingConcurrency")}</span><input className="input" type="number" min="1" step="1" value={pacingModelConcurrency} onChange={e => setPacingModelConcurrency(e.target.value)} /></label>
           <button type="button" className="btn btn-ghost btn-sm" onClick={addPacingModel}>{t("pws.pacingAdd")}</button>
         </div>
-        {Object.entries(pacingModels).length > 0 && <div className="pwi-pacing-overrides">{Object.entries(pacingModels).map(([model, rule]) => <div key={model} className="pwi-pacing-row"><code>{model}</code><span>{rule.requestsPerMinute !== undefined ? `${rule.requestsPerMinute} ${t("pws.pacingRpmUnit")}` : ""}{rule.requestsPerMinute !== undefined && rule.minIntervalMs !== undefined ? " · " : ""}{rule.minIntervalMs !== undefined ? `${rule.minIntervalMs} ms` : ""}</span><button type="button" className="btn btn-ghost btn-sm" onClick={() => setPacingModels(current => Object.fromEntries(Object.entries(current).filter(([id]) => id !== model)))} aria-label={t("pws.pacingRemoveModel", { model })}>{t("pws.pacingRemove")}</button></div>)}</div>}
+        {Object.entries(pacingModels).length > 0 && <div className="pwi-pacing-overrides">{Object.entries(pacingModels).map(([model, rule]) => <div key={model} className="pwi-pacing-row"><code>{model}</code><span>{[
+          rule.requestsPerMinute !== undefined ? `${rule.requestsPerMinute} ${t("pws.pacingRpmUnit")}` : undefined,
+          rule.minIntervalMs !== undefined ? `${rule.minIntervalMs} ms` : undefined,
+          rule.maxConcurrentRequests !== undefined ? `${rule.maxConcurrentRequests} ${t("pws.pacingCapUnit")}` : undefined,
+        ].filter(part => part !== undefined).join(" · ")}</span><button type="button" className="btn btn-ghost btn-sm" onClick={() => setPacingModels(current => Object.fromEntries(Object.entries(current).filter(([id]) => id !== model)))} aria-label={t("pws.pacingRemoveModel", { model })}>{t("pws.pacingRemove")}</button></div>)}</div>}
       </section>
       {formDirty && (
         <div className="pwi-settings-sticky-bar">

gui/src/i18n/de.ts

@@ -2688,9 +2688,11 @@ export const de: Record<TKey, string> = {
   "pws.pacingDesc": "Verteilt ausgehende Anfragestarts für diesen Anbieter gleichmäßig. Streaming-Antworten dürfen sich überlappen.",
   "pws.pacingEnabled": "Aktiviert",
   "pws.pacingRpm": "Anfragen pro Minute",
+  "pws.pacingConcurrency": "Maximale gleichzeitige Anfragen",
   "pws.pacingRpmUnit": "RPM",
+  "pws.pacingCapUnit": "×",
   "pws.pacingDelay": "Mindestintervall (ms)",
-  "pws.pacingSlowerWins": "Das langsamere Anbieterlimit gilt. Modellregeln können nur stärker verzögern.",
+  "pws.pacingSlowerWins": "Das langsamere Anbieterlimit gilt. Modellregeln können nur stärker verzögern oder die Zahl gleichzeitiger Anfragen senken.",
   "pws.pacingQueued": "in Warteschlange",
   "pws.pacingNextSlot": "bis zum nächsten Slot",
   "pws.pacingLastModel": "letztes Modell",
@@ -2701,6 +2703,7 @@ export const de: Record<TKey, string> = {
   "pws.pacingRemove": "Entfernen",
   "pws.pacingRemoveModel": "Anfragetaktung für {model} entfernen",
   "pws.pacingRuleRequired": "Legen Sie zuerst ein Anbieterlimit oder eine Modellregel fest.",
+  "pws.pacingCapInvalid": "Die maximale Anzahl gleichzeitiger Anfragen muss eine ganze Zahl größer oder gleich 1 sein.",
   "pws.saving": "Wird gespeichert…",
   "pws.settingsSaved": "Einstellungen gespeichert.",
   "pws.accountModeSaved": "Kontomodus gespeichert.",

gui/src/i18n/en.ts

@@ -1723,9 +1723,11 @@ export const en = {
   "pws.pacingDesc": "Evenly delay outbound request starts for this provider. Streaming responses may overlap.",
   "pws.pacingEnabled": "Enabled",
   "pws.pacingRpm": "Requests per minute",
+  "pws.pacingConcurrency": "Max concurrent requests",
   "pws.pacingRpmUnit": "RPM",
+  "pws.pacingCapUnit": "×",
   "pws.pacingDelay": "Minimum interval (ms)",
-  "pws.pacingSlowerWins": "The slower provider limit wins. Model overrides can only add more delay.",
+  "pws.pacingSlowerWins": "The slower provider limit wins. Model overrides can only add more delay or lower the concurrency cap.",
   "pws.pacingQueued": "queued",
   "pws.pacingNextSlot": "until next slot",
   "pws.pacingLastModel": "last model",
@@ -1736,6 +1738,7 @@ export const en = {
   "pws.pacingRemove": "Remove",
   "pws.pacingRemoveModel": "Remove request pacing override for {model}",
   "pws.pacingRuleRequired": "Enable request pacing only after setting a provider limit or a model override.",
+  "pws.pacingCapInvalid": "Max concurrent requests must be a whole number of 1 or more.",
   "pws.saving": "Saving…",
   "pws.settingsSaved": "Settings saved.",
   "pws.accountModeSaved": "Account mode saved.",

gui/src/i18n/fr.ts

@@ -1670,9 +1670,11 @@ export const fr: Record<TKey, string> = {
   "pws.pacingDesc": "Répartissez uniformément le démarrage des requêtes sortantes pour ce fournisseur. Les réponses diffusées peuvent se chevaucher.",
   "pws.pacingEnabled": "Activé",
   "pws.pacingRpm": "Requêtes par minute",
+  "pws.pacingConcurrency": "Requêtes simultanées max",
   "pws.pacingRpmUnit": "RPM",
+  "pws.pacingCapUnit": "×",
   "pws.pacingDelay": "Intervalle minimal (ms)",
-  "pws.pacingSlowerWins": "La limite la plus lente du fournisseur s’applique. Les remplacements par modèle ne peuvent qu’ajouter un délai.",
+  "pws.pacingSlowerWins": "La limite la plus lente du fournisseur s’applique. Les remplacements par modèle ne peuvent qu’ajouter un délai ou réduire le plafond de requêtes simultanées.",
   "pws.pacingQueued": "en attente",
   "pws.pacingNextSlot": "avant le prochain créneau",
   "pws.pacingLastModel": "dernier modèle",
@@ -1683,6 +1685,7 @@ export const fr: Record<TKey, string> = {
   "pws.pacingRemove": "Supprimer",
   "pws.pacingRemoveModel": "Supprimer le remplacement du cadencement des requêtes pour {model}",
   "pws.pacingRuleRequired": "Activez le cadencement des requêtes seulement après avoir défini une limite de fournisseur ou un remplacement par modèle.",
+  "pws.pacingCapInvalid": "Le nombre maximal de requêtes simultanées doit être un nombre entier supérieur ou égal à 1.",
   "pws.providerId": "ID du fournisseur",
   "pws.reauth": "Réauthentification requise",
   "pws.reauthenticate": "Réauthentifier",

gui/src/i18n/ja.ts

@@ -1599,9 +1599,11 @@ export const ja: Record<TKey, string> = {
   "pws.pacingDesc": "このプロバイダーへの送信開始を均等に遅延します。ストリーミング応答は重複できます。",
   "pws.pacingEnabled": "有効",
   "pws.pacingRpm": "1分あたりのリクエスト数",
+  "pws.pacingConcurrency": "最大同時リクエスト数",
   "pws.pacingRpmUnit": "RPM",
+  "pws.pacingCapUnit": "×",
   "pws.pacingDelay": "最小間隔 (ms)",
-  "pws.pacingSlowerWins": "より遅いプロバイダー制限が優先され、モデル設定は遅延を増やす場合のみ適用されます。",
+  "pws.pacingSlowerWins": "より遅いプロバイダー制限が優先され、モデル設定は遅延を増やすか同時リクエスト数の上限を下げる場合のみ適用されます。",
   "pws.pacingQueued": "待機中",
   "pws.pacingNextSlot": "次のスロットまで",
   "pws.pacingLastModel": "最後のモデル",
@@ -1612,6 +1614,7 @@ export const ja: Record<TKey, string> = {
   "pws.pacingRemove": "削除",
   "pws.pacingRemoveModel": "{model} のリクエスト間隔設定を削除",
   "pws.pacingRuleRequired": "プロバイダー制限またはモデル別設定を追加してから有効にしてください。",
+  "pws.pacingCapInvalid": "最大同時リクエスト数は 1 以上の整数にしてください。",
   "pws.saving": "保存中…",
   "pws.settingsSaved": "設定を保存しました。",
   "pws.accountModeSaved": "アカウントモードを保存しました。",

gui/src/i18n/ko.ts

@@ -2727,9 +2727,11 @@ export const ko: Record<TKey, string> = {
   "pws.pacingDesc": "이 프로바이더로 나가는 요청 시작을 일정 간격으로 지연합니다. 스트리밍 응답은 서로 겹칠 수 있습니다.",
   "pws.pacingEnabled": "사용",
   "pws.pacingRpm": "분당 요청 수",
+  "pws.pacingConcurrency": "최대 동시 요청 수",
   "pws.pacingRpmUnit": "RPM",
+  "pws.pacingCapUnit": "×",
   "pws.pacingDelay": "최소 시작 간격(ms)",
-  "pws.pacingSlowerWins": "프로바이더의 더 느린 제한이 우선합니다. 모델별 설정은 지연을 더 늘릴 때만 적용됩니다.",
+  "pws.pacingSlowerWins": "프로바이더의 더 느린 제한이 우선합니다. 모델별 설정은 지연을 더 늘리거나 동시 요청 상한을 낮출 때만 적용됩니다.",
   "pws.pacingQueued": "대기 중",
   "pws.pacingNextSlot": "다음 슬롯까지",
   "pws.pacingLastModel": "마지막 모델",
@@ -2740,6 +2742,7 @@ export const ko: Record<TKey, string> = {
   "pws.pacingRemove": "제거",
   "pws.pacingRemoveModel": "{model} 모델의 요청 속도 설정 제거",
   "pws.pacingRuleRequired": "프로바이더 제한이나 모델별 제한을 하나 이상 설정한 뒤 요청 속도 조절을 켜세요.",
+  "pws.pacingCapInvalid": "최대 동시 요청 수는 1 이상의 정수여야 합니다.",
   "pws.saving": "저장 중…",
   "pws.settingsSaved": "설정이 저장되었습니다.",
   "pws.accountModeSaved": "계정 모드가 저장되었습니다.",

gui/src/i18n/ru.ts

@@ -1690,9 +1690,11 @@ export const ru: Record<TKey, string> = {
   "pws.pacingDesc": "Равномерно задерживает начало исходящих запросов к провайдеру. Потоковые ответы могут пересекаться.",
   "pws.pacingEnabled": "Включено",
   "pws.pacingRpm": "Запросов в минуту",
+  "pws.pacingConcurrency": "Макс. одновременных запросов",
   "pws.pacingRpmUnit": "RPM",
+  "pws.pacingCapUnit": "×",
   "pws.pacingDelay": "Минимальный интервал (мс)",
-  "pws.pacingSlowerWins": "Действует более медленный лимит провайдера. Правила моделей могут только увеличить задержку.",
+  "pws.pacingSlowerWins": "Действует более медленный лимит провайдера. Правила моделей могут только увеличить задержку или снизить лимит одновременных запросов.",
   "pws.pacingQueued": "в очереди",
   "pws.pacingNextSlot": "до следующего слота",
   "pws.pacingLastModel": "последняя модель",
@@ -1703,6 +1705,7 @@ export const ru: Record<TKey, string> = {
   "pws.pacingRemove": "Удалить",
   "pws.pacingRemoveModel": "Удалить интервал запросов для {model}",
   "pws.pacingRuleRequired": "Сначала задайте лимит провайдера или правило модели.",
+  "pws.pacingCapInvalid": "Максимальное количество одновременных запросов должно быть целым числом не меньше 1.",
   "pws.saving": "Сохранение…",
   "pws.settingsSaved": "Настройки сохранены.",
   "pws.accountModeSaved": "Режим аккаунта сохранён.",