diff --git a/packages/types/src/providers/bedrock.ts b/packages/types/src/providers/bedrock.ts
index 62e180ad30..aa665ef1c4 100644
--- a/packages/types/src/providers/bedrock.ts
+++ b/packages/types/src/providers/bedrock.ts
@@ -173,6 +173,7 @@ export const bedrockModels = {
 		supportsImages: true,
 		supportsPromptCache: true,
 		supportsReasoningBudget: true,
+		supportsTemperature: false, // Temperature is deprecated for this model
 		inputPrice: 5.0, // $5 per million input tokens (≤200K context)
 		outputPrice: 25.0, // $25 per million output tokens (≤200K context)
 		cacheWritesPrice: 6.25, // $6.25 per million tokens
diff --git a/src/api/providers/bedrock.ts b/src/api/providers/bedrock.ts
index e91c1b3907..eb254cfa80 100644
--- a/src/api/providers/bedrock.ts
+++ b/src/api/providers/bedrock.ts
@@ -435,7 +435,10 @@ export class AwsBedrockHandler extends BaseProvider implements SingleCompletionH
 
 		const inferenceConfig: BedrockInferenceConfig = {
 			maxTokens: modelConfig.maxTokens || (modelConfig.info.maxTokens as number),
-			temperature: modelConfig.temperature ?? (this.options.modelTemperature as number),
+			// Only include temperature if the model supports it (Opus 4.7 deprecated temperature)
+			...(modelConfig.info.supportsTemperature !== false && {
+				temperature: modelConfig.temperature ?? (this.options.modelTemperature as number),
+			}),
 		}
 
 		// Check if 1M context is enabled for supported Claude 4 models
@@ -775,7 +778,10 @@ export class AwsBedrockHandler extends BaseProvider implements SingleCompletionH
 
 			const inferenceConfig: BedrockInferenceConfig = {
 				maxTokens: modelConfig.maxTokens || (modelConfig.info.maxTokens as number),
-				temperature: modelConfig.temperature ?? (this.options.modelTemperature as number),
+				// Only include temperature if the model supports it (Opus 4.7 deprecated temperature)
+				...(modelConfig.info.supportsTemperature !== false && {
+					temperature: modelConfig.temperature ?? (this.options.modelTemperature as number),
+				}),
 			}
 
 			// For completePrompt, use a unique conversation ID based on the prompt