fix: added WithUsePromptCaching,

NewHTTPProviderWithMaxTokensFieldAndRequestTimeoutAndUsePromptCaching,
used usePromptCaching for openai protocol only

pkg/providers/factory_provider.go

@@ -84,12 +84,14 @@ func CreateProviderFromConfig(cfg *config.ModelConfig) (LLMProvider, string, err
 		if apiBase == "" {
 			apiBase = getDefaultAPIBase(protocol)
 		}
-		return NewHTTPProviderWithMaxTokensFieldAndRequestTimeout(
+
+		return NewHTTPProviderWithMaxTokensFieldAndRequestTimeoutAndUsePromptCaching(
 			cfg.APIKey,
 			apiBase,
 			cfg.Proxy,
 			cfg.MaxTokensField,
 			cfg.RequestTimeout,
+			true,
 		), modelID, nil
 
 	case "openrouter", "groq", "zhipu", "gemini", "nvidia",
@@ -103,12 +105,13 @@ func CreateProviderFromConfig(cfg *config.ModelConfig) (LLMProvider, string, err
 		if apiBase == "" {
 			apiBase = getDefaultAPIBase(protocol)
 		}
-		return NewHTTPProviderWithMaxTokensFieldAndRequestTimeout(
+		return NewHTTPProviderWithMaxTokensFieldAndRequestTimeoutAndUsePromptCaching(
 			cfg.APIKey,
 			apiBase,
 			cfg.Proxy,
 			cfg.MaxTokensField,
 			cfg.RequestTimeout,
+			false,
 		), modelID, nil
 
 	case "anthropic":
@@ -128,12 +131,13 @@ func CreateProviderFromConfig(cfg *config.ModelConfig) (LLMProvider, string, err
 		if cfg.APIKey == "" {
 			return nil, "", fmt.Errorf("api_key is required for anthropic protocol (model: %s)", cfg.Model)
 		}
-		return NewHTTPProviderWithMaxTokensFieldAndRequestTimeout(
+		return NewHTTPProviderWithMaxTokensFieldAndRequestTimeoutAndUsePromptCaching(
 			cfg.APIKey,
 			apiBase,
 			cfg.Proxy,
 			cfg.MaxTokensField,
 			cfg.RequestTimeout,
+			false,
 		), modelID, nil
 
 	case "antigravity":

pkg/providers/http_provider.go

@@ -42,6 +42,23 @@ func NewHTTPProviderWithMaxTokensFieldAndRequestTimeout(
 	}
 }
 
+func NewHTTPProviderWithMaxTokensFieldAndRequestTimeoutAndUsePromptCaching(
+	apiKey, apiBase, proxy, maxTokensField string,
+	requestTimeoutSeconds int,
+	usePromptCaching bool,
+) *HTTPProvider {
+	return &HTTPProvider{
+		delegate: openai_compat.NewProvider(
+			apiKey,
+			apiBase,
+			proxy,
+			openai_compat.WithMaxTokensField(maxTokensField),
+			openai_compat.WithRequestTimeout(time.Duration(requestTimeoutSeconds)*time.Second),
+			openai_compat.WithUsePromptCaching(usePromptCaching),
+		),
+	}
+}
+
 func (p *HTTPProvider) Chat(
 	ctx context.Context,
 	messages []Message,

pkg/providers/openai_compat/provider.go

@@ -32,6 +32,7 @@ type Provider struct {
 	apiKey           string
 	apiBase          string
 	maxTokensField   string // Field name for max tokens (e.g., "max_completion_tokens" for o1/glm models)
+	usePromptCaching bool
 	httpClient       *http.Client
 }
 
@@ -53,6 +54,14 @@ func WithRequestTimeout(timeout time.Duration) Option {
 	}
 }
 
+// WithUsePromptCaching pass in true to enable prompt caching.
+// Prompt caching is only supported by OpenAI-native endpoints.
+func WithUsePromptCaching(usePromptCaching bool) Option {
+	return func(p *Provider) {
+		p.usePromptCaching = usePromptCaching
+	}
+}
+
 func NewProvider(apiKey, apiBase, proxy string, opts ...Option) *Provider {
 	client := &http.Client{
 		Timeout: defaultRequestTimeout,
@@ -157,7 +166,7 @@ func (p *Provider) Chat(
 	// Prompt caching is only supported by OpenAI-native endpoints.
 	// Gemini and other providers reject unknown fields, so skip for non-OpenAI APIs.
 	if cacheKey, ok := options["prompt_cache_key"].(string); ok && cacheKey != "" {
-		if !strings.Contains(p.apiBase, "generativelanguage.googleapis.com") {
+		if p.usePromptCaching {
 			requestBody["prompt_cache_key"] = cacheKey
 		}
 	}