diff --git a/manifests/models/deepseek-v4-flash.json b/manifests/models/deepseek-v4-flash.json
index b55e5716..877ef79e 100644
--- a/manifests/models/deepseek-v4-flash.json
+++ b/manifests/models/deepseek-v4-flash.json
@@ -2,40 +2,40 @@
   "$schema": "../$schemas/model.schema.json",
   "id": "deepseek-v4-flash",
   "name": "DeepSeek-V4-Flash",
-  "description": "DeepSeek's efficient open-weight V4 model with 284B total and 13B active parameters, 1M context, dual thinking modes, JSON output, and tool calls.",
+  "description": "DeepSeek's efficient open-weight V4 model, updated on July 31 with substantially enhanced agent capabilities while retaining 284B total and 13B active parameters.",
   "translations": {
     "de": {
-      "description": "DeepSeeks effizientes Open-Weight-V4-Modell mit 284 Mrd. Gesamt- und 13 Mrd. aktiven Parametern, 1 Mio. Kontext, zwei Denkmodi, JSON und Werkzeugaufrufen."
+      "description": "DeepSeeks effizientes Open-Weight-V4-Modell, das am 31. Juli mit deutlich verbesserten Agentenfähigkeiten aktualisiert wurde und weiterhin 284 Mrd. Gesamt- sowie 13 Mrd. aktive Parameter besitzt."
     },
     "es": {
-      "description": "El modelo V4 eficiente y de pesos abiertos de DeepSeek, con 284B parámetros totales y 13B activos, contexto de 1M, dos modos de pensamiento, JSON y herramientas."
+      "description": "El eficiente modelo V4 de pesos abiertos de DeepSeek, actualizado el 31 de julio con capacidades de agente considerablemente mejoradas y manteniendo 284B parámetros totales y 13B activos."
     },
     "fr": {
-      "description": "Le modèle V4 efficace à poids ouverts de DeepSeek, avec 284 Md de paramètres dont 13 Md actifs, contexte 1 M, deux modes de réflexion, JSON et outils."
+      "description": "Le modèle V4 efficace à poids ouverts de DeepSeek, mis à jour le 31 juillet avec des capacités agentiques nettement renforcées tout en conservant 284 Md de paramètres au total, dont 13 Md actifs."
     },
     "id": {
-      "description": "Model V4 open-weight efisien dari DeepSeek dengan 284B parameter total dan 13B aktif, konteks 1M, dua mode berpikir, keluaran JSON, dan pemanggilan alat."
+      "description": "Model V4 open-weight efisien dari DeepSeek, diperbarui pada 31 Juli dengan kemampuan agen yang jauh lebih kuat sambil mempertahankan 284B parameter total dan 13B parameter aktif."
     },
     "ja": {
-      "description": "総2840億・アクティブ130億パラメータ、100万コンテキスト、2種類の思考モード、JSON出力、ツール呼び出しを備えたDeepSeekの効率的なオープンウェイトV4モデル。"
+      "description": "総2840億・アクティブ130億パラメータを維持しつつ、7月31日にエージェント能力を大幅に強化したDeepSeekの効率的なオープンウェイトV4モデル。"
     },
     "ko": {
-      "description": "총 284B 및 활성 13B 매개변수, 100만 컨텍스트, 이중 사고 모드, JSON 출력과 도구 호출을 지원하는 DeepSeek의 효율적인 오픈 웨이트 V4 모델입니다."
+      "description": "총 284B 및 활성 13B 매개변수를 유지하면서 7월 31일 에이전트 기능을 크게 강화한 DeepSeek의 효율적인 오픈 웨이트 V4 모델입니다."
     },
     "pt": {
-      "description": "O modelo V4 eficiente e de pesos abertos da DeepSeek, com 284B parâmetros totais e 13B ativos, contexto de 1M, dois modos de pensamento, JSON e ferramentas."
+      "description": "O modelo V4 eficiente e de pesos abertos da DeepSeek, atualizado em 31 de julho com capacidades de agente significativamente aprimoradas, mantendo 284B parâmetros totais e 13B ativos."
     },
     "ru": {
-      "description": "Эффективная открытая модель DeepSeek V4: 284 млрд параметров, 13 млрд активных, контекст 1 млн, два режима рассуждений, JSON и вызов инструментов."
+      "description": "Эффективная модель DeepSeek V4 с открытыми весами, обновлённая 31 июля со значительно усиленными агентными возможностями при сохранении 284 млрд параметров, из которых 13 млрд активны."
     },
     "tr": {
-      "description": "DeepSeek'in 284B toplam ve 13B aktif parametreli, 1M bağlamlı, çift düşünme modlu, JSON çıktılı ve araç çağrılı verimli açık ağırlıklı V4 modeli."
+      "description": "DeepSeek'in 31 Temmuz'da aracı yetenekleri önemli ölçüde geliştirilerek güncellenen, 284B toplam ve 13B aktif parametreyi koruyan verimli açık ağırlıklı V4 modeli."
     },
     "zh-Hans": {
-      "description": "DeepSeek 的高效开放权重 V4 模型，2840 亿总参数、130 亿激活参数，支持 100 万上下文、双思考模式、JSON 输出和工具调用。"
+      "description": "DeepSeek 的高效开放权重 V4 模型，于 7 月 31 日更新并显著增强智能体能力，保持 2840 亿总参数和 130 亿激活参数。"
     },
     "zh-Hant": {
-      "description": "DeepSeek 的高效開放權重 V4 模型，2840 億總參數、130 億啟用參數，支援 100 萬上下文、雙思考模式、JSON 輸出和工具呼叫。"
+      "description": "DeepSeek 的高效開放權重 V4 模型，於 7 月 31 日更新並顯著增強智慧體能力，維持 2840 億總參數和 130 億啟用參數。"
     }
   },
   "verified": true,
@@ -43,15 +43,7 @@
     {
       "url": "https://api-docs.deepseek.com/news/news260424/",
       "title": "DeepSeek V4 Preview release",
-      "fields": [
-        "name",
-        "description",
-        "size",
-        "releaseDate",
-        "lifecycle",
-        "contextWindow",
-        "capabilities"
-      ]
+      "fields": ["name", "description", "size", "lifecycle", "contextWindow", "capabilities"]
     },
     {
       "url": "https://api-docs.deepseek.com/quick_start/pricing",
@@ -65,9 +57,19 @@
         "outputModalities",
         "capabilities"
       ]
+    },
+    {
+      "url": "https://api-docs.deepseek.com/updates/",
+      "title": "DeepSeek V4 Flash July 31 API update",
+      "fields": ["description", "releaseDate", "lifecycle", "capabilities"]
+    },
+    {
+      "url": "https://artificialanalysis.ai/models/deepseek-v4-flash/",
+      "title": "Artificial Analysis DeepSeek V4 Flash model page",
+      "fields": ["platformUrls.artificialAnalysis"]
     }
   ],
-  "lastVerifiedAt": "2026-07-21",
+  "lastVerifiedAt": "2026-07-31",
   "verifiedBy": "codex-agent",
   "confidence": "high",
   "websiteUrl": "https://www.deepseek.com",
@@ -101,7 +103,7 @@
       }
     ]
   },
-  "releaseDate": "2026-04-24",
+  "releaseDate": "2026-07-31",
   "lifecycle": "latest",
   "knowledgeCutoff": null,
   "inputModalities": ["text"],
@@ -118,7 +120,7 @@
   },
   "platformUrls": {
     "huggingface": "https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash",
-    "artificialAnalysis": null,
+    "artificialAnalysis": "https://artificialanalysis.ai/models/deepseek-v4-flash/",
     "openrouter": null
   }
 }
diff --git a/manifests/models/deepseek-v4-pro.json b/manifests/models/deepseek-v4-pro.json
index 7aee9fce..627d673a 100644
--- a/manifests/models/deepseek-v4-pro.json
+++ b/manifests/models/deepseek-v4-pro.json
@@ -1,7 +1,7 @@
 {
   "$schema": "../$schemas/model.schema.json",
   "id": "deepseek-v4-pro",
-  "name": "DeepSeek-V4-Pro",
+  "name": "DeepSeek-V4-Pro-Preview",
   "description": "DeepSeek's flagship open-weight V4 model with 1.6T total and 49B active parameters, 1M context, dual thinking modes, JSON output, and tool calls.",
   "translations": {
     "de": {
@@ -65,9 +65,14 @@
         "outputModalities",
         "capabilities"
       ]
+    },
+    {
+      "url": "https://api-docs.deepseek.com/updates/",
+      "title": "DeepSeek V4 Flash July 31 API update",
+      "fields": ["name", "lifecycle"]
     }
   ],
-  "lastVerifiedAt": "2026-07-21",
+  "lastVerifiedAt": "2026-07-31",
   "verifiedBy": "codex-agent",
   "confidence": "high",
   "websiteUrl": "https://www.deepseek.com",