diff --git a/packages/types/src/providers/deepseek.ts b/packages/types/src/providers/deepseek.ts
index 9387d6a4ae..d2bcd84607 100644
--- a/packages/types/src/providers/deepseek.ts
+++ b/packages/types/src/providers/deepseek.ts
@@ -14,7 +14,7 @@ export const deepSeekModels = {
 		contextWindow: 1_000_000,
 		supportsImages: true,
 		supportsPromptCache: true,
-		supportsReasoningEffort: ["disable", "low", "high", "max"], // Updated 2026-08-01
+		supportsReasoningEffort: ["disable", "low", "high", "max"], // Updated 2026-08-13
 		preserveReasoning: true,
 		reasoningEffort: "high",
 		inputPrice: 0, // the inputs are priced as cache read/write, so `inputPrice` should be 0
@@ -29,7 +29,7 @@ export const deepSeekModels = {
 		contextWindow: 1_000_000,
 		supportsImages: true,
 		supportsPromptCache: true,
-		supportsReasoningEffort: ["disable", "high", "max"], // Updated 2026-08-01
+		supportsReasoningEffort: ["disable", "low", "high", "max"], // Updated 2026-08-13
 		preserveReasoning: true,
 		reasoningEffort: "high",
 		inputPrice: 0, // the inputs are priced as cache read/write, so `inputPrice` should be 0
diff --git a/src/api/providers/__tests__/deepseek.spec.ts b/src/api/providers/__tests__/deepseek.spec.ts
index 4f3cccdc08..94bed4fc6f 100644
--- a/src/api/providers/__tests__/deepseek.spec.ts
+++ b/src/api/providers/__tests__/deepseek.spec.ts
@@ -617,6 +617,7 @@ describe("DeepSeekHandler", () => {
 
 	describe("normalizeDeepSeekReasoningEffort", () => {
 		// https://api-docs.deepseek.com/guides/thinking_mode/
+		// updated on 2026-08-13
 		it("should map acceptable reasoning efforts the same way as stated by the official documentation", async () => {
 			const mappings: {
 				modelId: DeepSeekModelId
@@ -633,6 +634,11 @@ describe("DeepSeekHandler", () => {
 					rawReasoningEffort: "low",
 					mappedReasoningEffort: "low",
 				},
+				{
+					modelId: "deepseek-v4-flash",
+					rawReasoningEffort: "medium",
+					mappedReasoningEffort: "high",
+				},
 				{
 					modelId: "deepseek-v4-flash",
 					rawReasoningEffort: "high",
@@ -656,6 +662,11 @@ describe("DeepSeekHandler", () => {
 				{
 					modelId: "deepseek-v4-pro",
 					rawReasoningEffort: "low",
+					mappedReasoningEffort: "low",
+				},
+				{
+					modelId: "deepseek-v4-pro",
+					rawReasoningEffort: "medium",
 					mappedReasoningEffort: "high",
 				},
 				{
@@ -666,7 +677,7 @@ describe("DeepSeekHandler", () => {
 				{
 					modelId: "deepseek-v4-pro",
 					rawReasoningEffort: "xhigh",
-					mappedReasoningEffort: "max",
+					mappedReasoningEffort: "high",
 				},
 				{
 					modelId: "deepseek-v4-pro",
diff --git a/src/api/providers/deepseek.ts b/src/api/providers/deepseek.ts
index 2e85c016b0..12d4afaae2 100644
--- a/src/api/providers/deepseek.ts
+++ b/src/api/providers/deepseek.ts
@@ -46,38 +46,22 @@ export const normalizeDeepSeekReasoningEffort = (
 	modelId: DeepSeekModelId,
 	reasoningEffort?: string,
 ): "low" | "high" | "max" | undefined => {
+	// still check the modelId so non-supported models won't produce reasoning efforts
 	switch (modelId) {
 		case "deepseek-v4-flash":
+		case "deepseek-v4-pro":
 			switch (reasoningEffort) {
 				case "low":
 					return "low"
 
+				case "medium":
 				case "high":
-					return "high"
-
 				case "xhigh":
 					return "high"
 
 				case "max":
 					return "max"
 			}
-			break
-
-		case "deepseek-v4-pro":
-			switch (reasoningEffort) {
-				case "low":
-					return "high"
-
-				case "high":
-					return "high"
-
-				case "xhigh":
-					return "max"
-
-				case "max":
-					return "max"
-			}
-			break
 	}
 
 	return undefined