Merge 9287b9276d into b867ec9145

src/api/providers/openai.ts

@@ -89,7 +89,20 @@ export class OpenAiHandler extends BaseProvider implements SingleCompletionHandl
 		const modelId = this.options.openAiModelId ?? ""
 		const enabledR1Format = this.options.openAiR1FormatEnabled ?? false
 		const isAzureAiInference = this._isAzureAiInference(modelUrl)
-		const deepseekReasoner = modelId.includes("deepseek-reasoner") || enabledR1Format
+		// Auto-detect reasoning/thinking models that require R1 format:
+		// - DeepSeek Reasoner models
+		// - Models with "thinking" suffix (e.g., kimi-k2-thinking)
+		// - DeepSeek R1 models (deepseek-r1, deepseek/deepseek-r1)
+		// - QWQ models
+		// - Or when user explicitly enables R1 format
+		const modelIdLower = modelId.toLowerCase()
+		const useR1Format =
+			enabledR1Format ||
+			modelIdLower.includes("deepseek-reasoner") ||
+			modelIdLower.includes("-thinking") ||
+			modelIdLower.includes("deepseek-r1") ||
+			modelIdLower.includes("/deepseek-r1") ||
+			modelIdLower.includes("qwq")
 
 		if (modelId.includes("o1") || modelId.includes("o3") || modelId.includes("o4")) {
 			yield* this.handleO3FamilyMessage(modelId, systemPrompt, messages, metadata)
@@ -104,7 +117,7 @@ export class OpenAiHandler extends BaseProvider implements SingleCompletionHandl
 		if (this.options.openAiStreamingEnabled ?? true) {
 			let convertedMessages
 
-			if (deepseekReasoner) {
+			if (useR1Format) {
 				convertedMessages = convertToR1Format([{ role: "user", content: systemPrompt }, ...messages])
 			} else {
 				if (modelInfo.supportsPromptCache) {
@@ -154,7 +167,7 @@ export class OpenAiHandler extends BaseProvider implements SingleCompletionHandl
 
 			const requestOptions: OpenAI.Chat.Completions.ChatCompletionCreateParamsStreaming = {
 				model: modelId,
-				temperature: this.options.modelTemperature ?? (deepseekReasoner ? DEEP_SEEK_DEFAULT_TEMPERATURE : 0),
+				temperature: this.options.modelTemperature ?? (useR1Format ? DEEP_SEEK_DEFAULT_TEMPERATURE : 0),
 				messages: convertedMessages,
 				stream: true as const,
 				...(isGrokXAI ? {} : { stream_options: { include_usage: true } }),
@@ -223,7 +236,7 @@ export class OpenAiHandler extends BaseProvider implements SingleCompletionHandl
 		} else {
 			const requestOptions: OpenAI.Chat.Completions.ChatCompletionCreateParamsNonStreaming = {
 				model: modelId,
-				messages: deepseekReasoner
+				messages: useR1Format
 					? convertToR1Format([{ role: "user", content: systemPrompt }, ...messages])
 					: [systemMessage, ...convertToOpenAiMessages(messages)],
 				// Tools are always present (minimum ALWAYS_AVAILABLE_TOOLS)

webview-ui/src/i18n/locales/ca/settings.json

@@ -834,8 +834,8 @@
 		"cacheReadsPrice": "Preu de lectures de caché",
 		"cacheWritesPrice": "Preu d'escriptures de caché",
 		"enableStreaming": "Habilitar streaming",
-		"enableR1Format": "Activar els paràmetres del model R1",
-		"enableR1FormatTips": "S'ha d'activat quan s'utilitzen models R1 com el QWQ per evitar errors 400",
+		"enableR1Format": "Activar format de model de raonament",
+		"enableR1FormatTips": "Activeu per a models de raonament/pensament (DeepSeek R1, Kimi-K2, QWQ, etc.) per gestionar correctament el format de resposta. Requerit per a models a NVIDIA NIM i altres proveïdors compatibles amb OpenAI.",
 		"useAzure": "Utilitzar Azure",
 		"azureApiVersion": "Establir versió de l'API d'Azure",
 		"gemini": {

webview-ui/src/i18n/locales/de/settings.json

@@ -834,8 +834,8 @@
 		"cacheReadsPrice": "Cache-Lesepreis",
 		"cacheWritesPrice": "Cache-Schreibpreis",
 		"enableStreaming": "Streaming aktivieren",
-		"enableR1Format": "R1-Modellparameter aktivieren",
-		"enableR1FormatTips": "Muss bei Verwendung von R1-Modellen wie QWQ aktiviert werden, um 400er-Fehler zu vermeiden",
+		"enableR1Format": "Reasoning-Modell-Format aktivieren",
+		"enableR1FormatTips": "Aktivieren für Reasoning-/Denkmodelle (DeepSeek R1, Kimi-K2, QWQ, usw.), um das Antwortformat korrekt zu verarbeiten. Erforderlich für Modelle auf NVIDIA NIM und anderen OpenAI-kompatiblen Anbietern.",
 		"useAzure": "Azure verwenden",
 		"azureApiVersion": "Azure API-Version festlegen",
 		"gemini": {

webview-ui/src/i18n/locales/en/settings.json

@@ -896,8 +896,8 @@
 		"cacheReadsPrice": "Cache reads price",
 		"cacheWritesPrice": "Cache writes price",
 		"enableStreaming": "Enable streaming",
-		"enableR1Format": "Enable R1 model parameters",
-		"enableR1FormatTips": "Must be enabled when using R1 models such as QWQ to prevent 400 errors",
+		"enableR1Format": "Enable reasoning model format",
+		"enableR1FormatTips": "Enable for reasoning/thinking models (DeepSeek R1, Kimi-K2, QWQ, etc.) to properly handle their response format. Required for models on NVIDIA NIM and other OpenAI-compatible providers.",
 		"useAzure": "Use Azure",
 		"azureApiVersion": "Set Azure API version",
 		"gemini": {

webview-ui/src/i18n/locales/es/settings.json

@@ -834,8 +834,8 @@
 		"cacheReadsPrice": "Precio de lecturas de caché",
 		"cacheWritesPrice": "Precio de escrituras de caché",
 		"enableStreaming": "Habilitar streaming",
-		"enableR1Format": "Habilitar parámetros del modelo R1",
-		"enableR1FormatTips": "Debe habilitarse al utilizar modelos R1 como QWQ, para evitar el error 400",
+		"enableR1Format": "Habilitar formato de modelo de razonamiento",
+		"enableR1FormatTips": "Habilitar para modelos de razonamiento/pensamiento (DeepSeek R1, Kimi-K2, QWQ, etc.) para manejar correctamente el formato de respuesta. Requerido para modelos en NVIDIA NIM y otros proveedores compatibles con OpenAI.",
 		"useAzure": "Usar Azure",
 		"azureApiVersion": "Establecer versión de API de Azure",
 		"gemini": {

webview-ui/src/i18n/locales/fr/settings.json

@@ -834,8 +834,8 @@
 		"cacheReadsPrice": "Prix des lectures de cache",
 		"cacheWritesPrice": "Prix des écritures de cache",
 		"enableStreaming": "Activer le streaming",
-		"enableR1Format": "Activer les paramètres du modèle R1",
-		"enableR1FormatTips": "Doit être activé lors de l'utilisation de modèles R1 tels que QWQ, pour éviter l'erreur 400",
+		"enableR1Format": "Activer le format de modèle de raisonnement",
+		"enableR1FormatTips": "Activer pour les modèles de raisonnement/thinking (DeepSeek R1, Kimi-K2, QWQ, etc.) pour gérer correctement le format des réponses. Requis pour les modèles sur NVIDIA NIM et autres fournisseurs compatibles OpenAI.",
 		"useAzure": "Utiliser Azure",
 		"azureApiVersion": "Définir la version de l'API Azure",
 		"gemini": {

webview-ui/src/i18n/locales/hi/settings.json

@@ -834,8 +834,8 @@
 		"cacheReadsPrice": "कैश रीड्स मूल्य",
 		"cacheWritesPrice": "कैश राइट्स मूल्य",
 		"enableStreaming": "स्ट्रीमिंग सक्षम करें",
-		"enableR1Format": "R1 मॉडल पैरामीटर सक्षम करें",
-		"enableR1FormatTips": "QWQ जैसी R1 मॉडलों का उपयोग करते समय इसे सक्षम करना आवश्यक है, ताकि 400 त्रुटि से बचा जा सके",
+		"enableR1Format": "रीजनिंग मॉडल फॉर्मेट सक्षम करें",
+		"enableR1FormatTips": "रीजनिंग/थिंकिंग मॉडल (DeepSeek R1, Kimi-K2, QWQ, आदि) के लिए सक्षम करें ताकि उनके रिस्पॉन्स फॉर्मेट को सही ढंग से हैंडल किया जा सके। NVIDIA NIM और अन्य OpenAI-संगत प्रोवाइडर्स पर मॉडल के लिए आवश्यक।",
 		"useAzure": "Azure का उपयोग करें",
 		"azureApiVersion": "Azure API संस्करण सेट करें",
 		"gemini": {

webview-ui/src/i18n/locales/id/settings.json

@@ -834,8 +834,8 @@
 		"cacheReadsPrice": "Harga cache reads",
 		"cacheWritesPrice": "Harga cache writes",
 		"enableStreaming": "Aktifkan streaming",
-		"enableR1Format": "Aktifkan parameter model R1",
-		"enableR1FormatTips": "Harus diaktifkan saat menggunakan model R1 seperti QWQ untuk mencegah error 400",
+		"enableR1Format": "Aktifkan format model reasoning",
+		"enableR1FormatTips": "Aktifkan untuk model reasoning/thinking (DeepSeek R1, Kimi-K2, QWQ, dll.) untuk menangani format respons dengan benar. Diperlukan untuk model di NVIDIA NIM dan penyedia lain yang kompatibel dengan OpenAI.",
 		"useAzure": "Gunakan Azure",
 		"azureApiVersion": "Atur versi API Azure",
 		"gemini": {

webview-ui/src/i18n/locales/it/settings.json

@@ -834,8 +834,8 @@
 		"cacheReadsPrice": "Prezzo letture cache",
 		"cacheWritesPrice": "Prezzo scritture cache",
 		"enableStreaming": "Abilita streaming",
-		"enableR1Format": "Abilita i parametri del modello R1",
-		"enableR1FormatTips": "Deve essere abilitato quando si utilizzano modelli R1 come QWQ, per evitare l'errore 400",
+		"enableR1Format": "Abilita formato modello di ragionamento",
+		"enableR1FormatTips": "Abilita per modelli di ragionamento/thinking (DeepSeek R1, Kimi-K2, QWQ, ecc.) per gestire correttamente il formato delle risposte. Richiesto per modelli su NVIDIA NIM e altri provider compatibili con OpenAI.",
 		"useAzure": "Usa Azure",
 		"azureApiVersion": "Imposta versione API Azure",
 		"gemini": {

webview-ui/src/i18n/locales/ja/settings.json

@@ -834,8 +834,8 @@
 		"cacheReadsPrice": "キャッシュ読み取り価格",
 		"cacheWritesPrice": "キャッシュ書き込み価格",
 		"enableStreaming": "ストリーミングを有効化",
-		"enableR1Format": "R1モデルパラメータを有効にする",
-		"enableR1FormatTips": "QWQなどのR1モデルを使用する際には、有効にする必要があります。400エラーを防ぐために",
+		"enableR1Format": "推論モデルフォーマットを有効にする",
+		"enableR1FormatTips": "推論/思考モデル（DeepSeek R1、Kimi-K2、QWQ など）のレスポンスフォーマットを正しく処理するために有効にしてください。NVIDIA NIM およびその他の OpenAI 互換プロバイダーのモデルに必要です。",
 		"useAzure": "Azureを使用",
 		"azureApiVersion": "Azure APIバージョンを設定",
 		"gemini": {

webview-ui/src/i18n/locales/ko/settings.json

@@ -834,8 +834,8 @@
 		"cacheReadsPrice": "캐시 읽기 가격",
 		"cacheWritesPrice": "캐시 쓰기 가격",
 		"enableStreaming": "스트리밍 활성화",
-		"enableR1Format": "R1 모델 매개변수 활성화",
-		"enableR1FormatTips": "QWQ와 같은 R1 모델을 사용할 때 활성화해야 하며, 400 오류를 방지합니다",
+		"enableR1Format": "추론 모델 형식 활성화",
+		"enableR1FormatTips": "추론/사고 모델(DeepSeek R1, Kimi-K2, QWQ 등)의 응답 형식을 올바르게 처리하려면 활성화하세요. NVIDIA NIM 및 기타 OpenAI 호환 제공업체의 모델에 필요합니다.",
 		"useAzure": "Azure 사용",
 		"azureApiVersion": "Azure API 버전 설정",
 		"gemini": {

webview-ui/src/i18n/locales/nl/settings.json

@@ -834,8 +834,8 @@
 		"cacheReadsPrice": "Cache-leesprijs",
 		"cacheWritesPrice": "Cache-schrijfprijs",
 		"enableStreaming": "Streaming inschakelen",
-		"enableR1Format": "R1-modelparameters inschakelen",
-		"enableR1FormatTips": "Moet ingeschakeld zijn bij gebruik van R1-modellen zoals QWQ om 400-fouten te voorkomen",
+		"enableR1Format": "Redeneermodel-formaat inschakelen",
+		"enableR1FormatTips": "Inschakelen voor redeneer-/denkmodellen (DeepSeek R1, Kimi-K2, QWQ, enz.) om het responsformaat correct te verwerken. Vereist voor modellen op NVIDIA NIM en andere OpenAI-compatibele providers.",
 		"useAzure": "Azure gebruiken",
 		"azureApiVersion": "Azure API-versie instellen",
 		"gemini": {

webview-ui/src/i18n/locales/pl/settings.json

@@ -834,8 +834,8 @@
 		"cacheReadsPrice": "Cena odczytów bufora",
 		"cacheWritesPrice": "Cena zapisów bufora",
 		"enableStreaming": "Włącz strumieniowanie",
-		"enableR1Format": "Włącz parametry modelu R1",
-		"enableR1FormatTips": "Należy włączyć podczas korzystania z modeli R1, takich jak QWQ, aby uniknąć błędu 400",
+		"enableR1Format": "Włącz format modelu rozumowania",
+		"enableR1FormatTips": "Włącz dla modeli rozumowania/thinking (DeepSeek R1, Kimi-K2, QWQ, itp.) aby prawidłowo obsługiwać format odpowiedzi. Wymagane dla modeli na NVIDIA NIM i innych dostawcach zgodnych z OpenAI.",
 		"useAzure": "Użyj Azure",
 		"azureApiVersion": "Ustaw wersję API Azure",
 		"gemini": {

webview-ui/src/i18n/locales/pt-BR/settings.json

@@ -834,8 +834,8 @@
 		"cacheReadsPrice": "Preço de leituras de cache",
 		"cacheWritesPrice": "Preço de escritas de cache",
 		"enableStreaming": "Ativar streaming",
-		"enableR1Format": "Ativar parâmetros do modelo R1",
-		"enableR1FormatTips": "Deve ser ativado ao usar modelos R1 como QWQ, para evitar erro 400",
+		"enableR1Format": "Ativar formato de modelo de raciocínio",
+		"enableR1FormatTips": "Ative para modelos de raciocínio/pensamento (DeepSeek R1, Kimi-K2, QWQ, etc.) para tratar corretamente o formato de resposta. Necessário para modelos no NVIDIA NIM e outros provedores compatíveis com OpenAI.",
 		"useAzure": "Usar Azure",
 		"azureApiVersion": "Definir versão da API Azure",
 		"gemini": {

webview-ui/src/i18n/locales/ru/settings.json

@@ -834,8 +834,8 @@
 		"cacheReadsPrice": "Цена чтения из кэша",
 		"cacheWritesPrice": "Цена записи в кэш",
 		"enableStreaming": "Включить потоковую передачу",
-		"enableR1Format": "Включить параметры модели R1",
-		"enableR1FormatTips": "Необходимо включить при использовании моделей R1 (например, QWQ), чтобы избежать ошибок 400",
+		"enableR1Format": "Включить формат модели рассуждения",
+		"enableR1FormatTips": "Включите для моделей рассуждения/мышления (DeepSeek R1, Kimi-K2, QWQ и др.) для правильной обработки формата ответа. Требуется для моделей на NVIDIA NIM и других провайдерах, совместимых с OpenAI.",
 		"useAzure": "Использовать Azure",
 		"azureApiVersion": "Установить версию API Azure",
 		"gemini": {

webview-ui/src/i18n/locales/tr/settings.json

@@ -834,8 +834,8 @@
 		"cacheReadsPrice": "Önbellek okuma fiyatı",
 		"cacheWritesPrice": "Önbellek yazma fiyatı",
 		"enableStreaming": "Akışı etkinleştir",
-		"enableR1Format": "R1 model parametrelerini etkinleştir",
-		"enableR1FormatTips": "QWQ gibi R1 modelleri kullanıldığında etkinleştirilmelidir, 400 hatası alınmaması için",
+		"enableR1Format": "Akıl yürütme modeli formatını etkinleştir",
+		"enableR1FormatTips": "Yanıt formatını doğru şekilde işlemek için akıl yürütme/düşünme modelleri (DeepSeek R1, Kimi-K2, QWQ, vb.) için etkinleştirin. NVIDIA NIM ve diğer OpenAI uyumlu sağlayıcılardaki modeller için gereklidir.",
 		"useAzure": "Azure kullan",
 		"azureApiVersion": "Azure API sürümünü ayarla",
 		"gemini": {

webview-ui/src/i18n/locales/vi/settings.json

@@ -834,8 +834,8 @@
 		"cacheReadsPrice": "Giá đọc bộ nhớ đệm",
 		"cacheWritesPrice": "Giá ghi bộ nhớ đệm",
 		"enableStreaming": "Bật streaming",
-		"enableR1Format": "Kích hoạt tham số mô hình R1",
-		"enableR1FormatTips": "Cần kích hoạt khi sử dụng các mô hình R1 như QWQ, để tránh lỗi 400",
+		"enableR1Format": "Kích hoạt định dạng mô hình suy luận",
+		"enableR1FormatTips": "Kích hoạt cho các mô hình suy luận/thinking (DeepSeek R1, Kimi-K2, QWQ, v.v.) để xử lý đúng định dạng phản hồi. Cần thiết cho các mô hình trên NVIDIA NIM và các provider tương thích OpenAI khác.",
 		"useAzure": "Sử dụng Azure",
 		"azureApiVersion": "Đặt phiên bản API Azure",
 		"gemini": {

webview-ui/src/i18n/locales/zh-CN/settings.json

@@ -834,8 +834,8 @@
 		"cacheReadsPrice": "缓存读取价格",
 		"cacheWritesPrice": "缓存写入价格",
 		"enableStreaming": "启用流式传输",
-		"enableR1Format": "启用 R1 模型参数",
-		"enableR1FormatTips": "使用 QWQ 等 R1 系列模型时必须启用，避免出现 400 错误",
+		"enableR1Format": "启用推理模型格式",
+		"enableR1FormatTips": "为推理/思考模型（DeepSeek R1、Kimi-K2、QWQ 等）启用，以正确处理响应格式。在 NVIDIA NIM 和其他 OpenAI 兼容提供商上使用模型时需要启用。",
 		"useAzure": "使用 Azure 服务",
 		"azureApiVersion": "设置 Azure API 版本",
 		"gemini": {

webview-ui/src/i18n/locales/zh-TW/settings.json

@@ -844,8 +844,8 @@
 		"cacheReadsPrice": "快取讀取價格",
 		"cacheWritesPrice": "快取寫入價格",
 		"enableStreaming": "啟用串流輸出",
-		"enableR1Format": "啟用 R1 模型參數",
-		"enableR1FormatTips": "使用 QWQ 等 R1 模型時必須啟用，以避免發生 400 錯誤",
+		"enableR1Format": "啟用推理模型格式",
+		"enableR1FormatTips": "為推理/思考模型（DeepSeek R1、Kimi-K2、QWQ 等）啟用，以正確處理回應格式。在 NVIDIA NIM 和其他 OpenAI 相容提供者上使用模型時需要啟用。",
 		"useAzure": "使用 Azure",
 		"azureApiVersion": "設定 Azure API 版本",
 		"gemini": {