@ax-llm/ax 21.0.4 → 21.0.5

package/index.d.cts

@@ -2427,6 +2427,7 @@ declare class AxBaseAI<TModel, TEmbedModel, TChatRequest, TEmbedRequest, TChatRe
     private calculatePercentile;
     private updateLatencyMetrics;
     private updateErrorMetrics;
+    private recordEstimatedCost;
     private recordTokenUsage;
     private calculateRequestSize;
     private calculateResponseSize;
@@ -2535,6 +2536,9 @@ type AxModelInfo = {
     maxTokens?: number;
     isExpensive?: boolean;
     contextWindow?: number;
+    isDeprecated?: boolean;
+    /** ISO date (YYYY-MM-DD) the upstream provider will stop serving this model. */
+    deprecatedOn?: string;
 };
 type AxTokenUsage = {
     promptTokens: number;
@@ -2755,6 +2759,10 @@ type AxCitation = {
     license?: string;
     publicationDate?: string;
     snippet?: string;
+    /** File Search multimodal (Gemini, May 2026): media chunk id. */
+    mediaId?: string;
+    /** File Search multimodal (Gemini, May 2026): page numbers cited within the source. */
+    pageNumbers?: number[];
 };
 type AxModelUsage = {
     ai: string;
@@ -5709,6 +5717,10 @@ type AxAIAnthropicThinkingWire = {
 type AxAIAnthropicEffortLevel = 'low' | 'medium' | 'high' | 'max';
 type AxAIAnthropicOutputConfig = {
     effort?: AxAIAnthropicEffortLevel;
+    format?: {
+        type: 'json_schema';
+        schema: object;
+    };
 };
 type AxAIAnthropicEffortLevelMapping = {
     minimal?: AxAIAnthropicEffortLevel;
@@ -5771,7 +5783,7 @@ type AxAIAnthropicChatRequest = {
                 media_type: string;
                 data: string;
             };
-        } & AxAIAnthropicChatRequestCacheParam) | {
+        } & AxAIAnthropicChatRequestCacheParam) | ({
             type: 'tool_result';
             is_error?: boolean;
             tool_use_id: string;
@@ -5786,7 +5798,7 @@ type AxAIAnthropicChatRequest = {
                     data: string;
                 };
             } & AxAIAnthropicChatRequestCacheParam))[];
-        })[];
+        } & AxAIAnthropicChatRequestCacheParam))[];
     } | {
         role: 'assistant';
         content: string | (({
@@ -5826,10 +5838,6 @@ type AxAIAnthropicChatRequest = {
     top_k?: number;
     thinking?: AxAIAnthropicThinkingWire;
     output_config?: AxAIAnthropicOutputConfig;
-    output_format?: {
-        type: 'json_schema';
-        schema: object;
-    };
     metadata?: {
         user_id: string;
     };
@@ -6003,8 +6011,11 @@ declare enum AxAIOpenAIModel {
     GPT4OMini = "gpt-4o-mini",
     GPTAudio = "gpt-audio",
     GPTAudioMini = "gpt-audio-mini",
+    GPTAudio15 = "gpt-audio-1.5",
+    GPTRealtime15 = "gpt-realtime-1.5",
     GPTRealtime2 = "gpt-realtime-2",
     GPTRealtimeWhisper = "gpt-realtime-whisper",
+    GPTRealtimeTranslate = "gpt-realtime-translate",
     GPT4ChatGPT4O = "chatgpt-4o-latest",
     GPT4Turbo = "gpt-4-turbo",
     GPT35Turbo = "gpt-3.5-turbo",
@@ -6031,6 +6042,8 @@ declare enum AxAIOpenAIModel {
     GPT54 = "gpt-5.4",
     GPT54Mini = "gpt-5.4-mini",
     GPT54Nano = "gpt-5.4-nano",
+    GPT55 = "gpt-5.5",
+    GPT55Pro = "gpt-5.5-pro",
     O1 = "o1",
     O1Mini = "o1-mini",
     O3 = "o3",
@@ -6063,7 +6076,7 @@ type AxAIOpenAIConfig<TModel, TEmbedModel> = Omit<AxModelConfig, 'topK'> & {
     logprobs?: number;
     echo?: boolean;
     dimensions?: number;
-    reasoningEffort?: 'minimal' | 'low' | 'medium' | 'high';
+    reasoningEffort?: 'none' | 'minimal' | 'low' | 'medium' | 'high' | 'xhigh';
     store?: boolean;
     serviceTier?: 'auto' | 'default' | 'flex';
     webSearchOptions?: {
@@ -6111,7 +6124,7 @@ interface AxAIOpenAIResponseDelta<T> {
 }
 type AxAIOpenAIChatRequest<TModel> = {
     model: TModel;
-    reasoning_effort?: 'none' | 'minimal' | 'low' | 'medium' | 'high';
+    reasoning_effort?: 'none' | 'minimal' | 'low' | 'medium' | 'high' | 'xhigh';
     store?: boolean;
     modalities?: readonly ('text' | 'audio')[];
     audio?: {
@@ -6946,6 +6959,10 @@ declare enum AxAIGoogleGeminiModel {
     Gemini3Flash = "gemini-3-flash-preview",
     Gemini3Pro = "gemini-3.1-pro-preview",
     Gemini3ProImage = "gemini-3-pro-image-preview",
+    Gemini31FlashImage = "gemini-3.1-flash-image-preview",
+    Gemini31FlashTTS = "gemini-3.1-flash-tts-preview",
+    NanoBanana2 = "nano-banana-2",
+    GeminiRoboticsER16 = "gemini-robotics-er-1.6-preview",
     Gemini25Pro = "gemini-2.5-pro",
     Gemini25Flash = "gemini-2.5-flash",
     Gemini25FlashNativeAudio = "gemini-2.5-flash-native-audio-preview-12-2025",
@@ -6967,7 +6984,6 @@ declare enum AxAIGoogleGeminiEmbedModel {
     GeminiEmbedding001 = "gemini-embedding-001",
     GeminiEmbedding = "gemini-embedding-exp",
     TextEmbeddingLarge = "text-embedding-large-exp-03-07",
-    TextEmbedding004 = "text-embedding-004",
     TextEmbedding005 = "text-embedding-005"
 }
 declare enum AxAIGoogleGeminiSafetyCategory {
@@ -7123,6 +7139,14 @@ type AxAIGoogleGeminiChatResponse = {
                     title?: string;
                     uri?: string;
                 };
+                retrievedContext?: {
+                    title?: string;
+                    uri?: string;
+                    /** File Search multimodal (May 2026): id of the file/media chunk. */
+                    media_id?: string;
+                    /** File Search multimodal (May 2026): page numbers cited within the source. */
+                    page_numbers?: number[];
+                };
             }[];
             googleMapsWidgetContextToken?: string;
         };
@@ -7527,6 +7551,8 @@ declare enum AxAIOpenAIResponsesModel {
     GPT54 = "gpt-5.4",
     GPT54Mini = "gpt-5.4-mini",
     GPT54Nano = "gpt-5.4-nano",
+    GPT55 = "gpt-5.5",
+    GPT55Pro = "gpt-5.5-pro",
     O1Pro = "o1-pro",
     O1 = "o1",
     O1Mini = "o1-mini",
@@ -7592,14 +7618,14 @@ interface AxAIOpenAIResponsesRequest<TModel = AxAIOpenAIResponsesModel> {
     readonly input: string | ReadonlyArray<AxAIOpenAIResponsesInputItem>;
     readonly model: TModel;
     readonly background?: boolean | null;
-    readonly include?: ReadonlyArray<'file_search_call.results' | 'message.input_image.image_url' | 'computer_call_output.output.image_url' | 'reasoning.encrypted_content' | 'code_interpreter_call.outputs'> | null;
+    readonly include?: ReadonlyArray<'file_search_call.results' | 'message.input_image.image_url' | 'computer_call_output.output.image_url' | 'reasoning.encrypted_content' | 'code_interpreter_call.outputs' | 'web_search_call.action.return_token_budget'> | null;
     readonly instructions?: string | null;
     readonly max_output_tokens?: number | null;
     readonly metadata?: Readonly<Record<string, string>> | null;
     readonly parallel_tool_calls?: boolean | null;
     readonly previous_response_id?: string | null;
     readonly reasoning?: {
-        readonly effort?: 'minimal' | 'low' | 'medium' | 'high' | null;
+        readonly effort?: 'none' | 'minimal' | 'low' | 'medium' | 'high' | 'xhigh' | null;
         readonly summary?: 'auto' | 'concise' | 'detailed' | null;
     } | null;
     readonly service_tier?: 'auto' | 'default' | 'flex' | null;
@@ -7629,6 +7655,7 @@ interface AxAIOpenAIResponsesOutputMessageItem {
     role: 'assistant';
     content: ReadonlyArray<AxAIOpenAIResponsesOutputTextContentPart | AxAIOpenAIResponsesOutputRefusalContentPart>;
     status: 'in_progress' | 'completed' | 'incomplete';
+    phase?: 'commentary' | 'final_answer';
 }
 interface AxAIOpenAIResponsesFunctionCallItem {
     type: 'function_call';
@@ -7978,7 +8005,7 @@ type AxAIOpenAIResponsesConfig<TModel, TEmbedModel> = Omit<AxModelConfig, 'topK'
     logprobs?: number;
     echo?: boolean;
     dimensions?: number;
-    reasoningEffort?: 'minimal' | 'low' | 'medium' | 'high';
+    reasoningEffort?: 'none' | 'minimal' | 'low' | 'medium' | 'high' | 'xhigh';
     reasoningSummary?: 'auto' | 'concise' | 'detailed';
     store?: boolean;
     systemPrompt?: string;

package/index.d.ts

@@ -2427,6 +2427,7 @@ declare class AxBaseAI<TModel, TEmbedModel, TChatRequest, TEmbedRequest, TChatRe
     private calculatePercentile;
     private updateLatencyMetrics;
     private updateErrorMetrics;
+    private recordEstimatedCost;
     private recordTokenUsage;
     private calculateRequestSize;
     private calculateResponseSize;
@@ -2535,6 +2536,9 @@ type AxModelInfo = {
     maxTokens?: number;
     isExpensive?: boolean;
     contextWindow?: number;
+    isDeprecated?: boolean;
+    /** ISO date (YYYY-MM-DD) the upstream provider will stop serving this model. */
+    deprecatedOn?: string;
 };
 type AxTokenUsage = {
     promptTokens: number;
@@ -2755,6 +2759,10 @@ type AxCitation = {
     license?: string;
     publicationDate?: string;
     snippet?: string;
+    /** File Search multimodal (Gemini, May 2026): media chunk id. */
+    mediaId?: string;
+    /** File Search multimodal (Gemini, May 2026): page numbers cited within the source. */
+    pageNumbers?: number[];
 };
 type AxModelUsage = {
     ai: string;
@@ -5709,6 +5717,10 @@ type AxAIAnthropicThinkingWire = {
 type AxAIAnthropicEffortLevel = 'low' | 'medium' | 'high' | 'max';
 type AxAIAnthropicOutputConfig = {
     effort?: AxAIAnthropicEffortLevel;
+    format?: {
+        type: 'json_schema';
+        schema: object;
+    };
 };
 type AxAIAnthropicEffortLevelMapping = {
     minimal?: AxAIAnthropicEffortLevel;
@@ -5771,7 +5783,7 @@ type AxAIAnthropicChatRequest = {
                 media_type: string;
                 data: string;
             };
-        } & AxAIAnthropicChatRequestCacheParam) | {
+        } & AxAIAnthropicChatRequestCacheParam) | ({
             type: 'tool_result';
             is_error?: boolean;
             tool_use_id: string;
@@ -5786,7 +5798,7 @@ type AxAIAnthropicChatRequest = {
                     data: string;
                 };
             } & AxAIAnthropicChatRequestCacheParam))[];
-        })[];
+        } & AxAIAnthropicChatRequestCacheParam))[];
     } | {
         role: 'assistant';
         content: string | (({
@@ -5826,10 +5838,6 @@ type AxAIAnthropicChatRequest = {
     top_k?: number;
     thinking?: AxAIAnthropicThinkingWire;
     output_config?: AxAIAnthropicOutputConfig;
-    output_format?: {
-        type: 'json_schema';
-        schema: object;
-    };
     metadata?: {
         user_id: string;
     };
@@ -6003,8 +6011,11 @@ declare enum AxAIOpenAIModel {
     GPT4OMini = "gpt-4o-mini",
     GPTAudio = "gpt-audio",
     GPTAudioMini = "gpt-audio-mini",
+    GPTAudio15 = "gpt-audio-1.5",
+    GPTRealtime15 = "gpt-realtime-1.5",
     GPTRealtime2 = "gpt-realtime-2",
     GPTRealtimeWhisper = "gpt-realtime-whisper",
+    GPTRealtimeTranslate = "gpt-realtime-translate",
     GPT4ChatGPT4O = "chatgpt-4o-latest",
     GPT4Turbo = "gpt-4-turbo",
     GPT35Turbo = "gpt-3.5-turbo",
@@ -6031,6 +6042,8 @@ declare enum AxAIOpenAIModel {
     GPT54 = "gpt-5.4",
     GPT54Mini = "gpt-5.4-mini",
     GPT54Nano = "gpt-5.4-nano",
+    GPT55 = "gpt-5.5",
+    GPT55Pro = "gpt-5.5-pro",
     O1 = "o1",
     O1Mini = "o1-mini",
     O3 = "o3",
@@ -6063,7 +6076,7 @@ type AxAIOpenAIConfig<TModel, TEmbedModel> = Omit<AxModelConfig, 'topK'> & {
     logprobs?: number;
     echo?: boolean;
     dimensions?: number;
-    reasoningEffort?: 'minimal' | 'low' | 'medium' | 'high';
+    reasoningEffort?: 'none' | 'minimal' | 'low' | 'medium' | 'high' | 'xhigh';
     store?: boolean;
     serviceTier?: 'auto' | 'default' | 'flex';
     webSearchOptions?: {
@@ -6111,7 +6124,7 @@ interface AxAIOpenAIResponseDelta<T> {
 }
 type AxAIOpenAIChatRequest<TModel> = {
     model: TModel;
-    reasoning_effort?: 'none' | 'minimal' | 'low' | 'medium' | 'high';
+    reasoning_effort?: 'none' | 'minimal' | 'low' | 'medium' | 'high' | 'xhigh';
     store?: boolean;
     modalities?: readonly ('text' | 'audio')[];
     audio?: {
@@ -6946,6 +6959,10 @@ declare enum AxAIGoogleGeminiModel {
     Gemini3Flash = "gemini-3-flash-preview",
     Gemini3Pro = "gemini-3.1-pro-preview",
     Gemini3ProImage = "gemini-3-pro-image-preview",
+    Gemini31FlashImage = "gemini-3.1-flash-image-preview",
+    Gemini31FlashTTS = "gemini-3.1-flash-tts-preview",
+    NanoBanana2 = "nano-banana-2",
+    GeminiRoboticsER16 = "gemini-robotics-er-1.6-preview",
     Gemini25Pro = "gemini-2.5-pro",
     Gemini25Flash = "gemini-2.5-flash",
     Gemini25FlashNativeAudio = "gemini-2.5-flash-native-audio-preview-12-2025",
@@ -6967,7 +6984,6 @@ declare enum AxAIGoogleGeminiEmbedModel {
     GeminiEmbedding001 = "gemini-embedding-001",
     GeminiEmbedding = "gemini-embedding-exp",
     TextEmbeddingLarge = "text-embedding-large-exp-03-07",
-    TextEmbedding004 = "text-embedding-004",
     TextEmbedding005 = "text-embedding-005"
 }
 declare enum AxAIGoogleGeminiSafetyCategory {
@@ -7123,6 +7139,14 @@ type AxAIGoogleGeminiChatResponse = {
                     title?: string;
                     uri?: string;
                 };
+                retrievedContext?: {
+                    title?: string;
+                    uri?: string;
+                    /** File Search multimodal (May 2026): id of the file/media chunk. */
+                    media_id?: string;
+                    /** File Search multimodal (May 2026): page numbers cited within the source. */
+                    page_numbers?: number[];
+                };
             }[];
             googleMapsWidgetContextToken?: string;
         };
@@ -7527,6 +7551,8 @@ declare enum AxAIOpenAIResponsesModel {
     GPT54 = "gpt-5.4",
     GPT54Mini = "gpt-5.4-mini",
     GPT54Nano = "gpt-5.4-nano",
+    GPT55 = "gpt-5.5",
+    GPT55Pro = "gpt-5.5-pro",
     O1Pro = "o1-pro",
     O1 = "o1",
     O1Mini = "o1-mini",
@@ -7592,14 +7618,14 @@ interface AxAIOpenAIResponsesRequest<TModel = AxAIOpenAIResponsesModel> {
     readonly input: string | ReadonlyArray<AxAIOpenAIResponsesInputItem>;
     readonly model: TModel;
     readonly background?: boolean | null;
-    readonly include?: ReadonlyArray<'file_search_call.results' | 'message.input_image.image_url' | 'computer_call_output.output.image_url' | 'reasoning.encrypted_content' | 'code_interpreter_call.outputs'> | null;
+    readonly include?: ReadonlyArray<'file_search_call.results' | 'message.input_image.image_url' | 'computer_call_output.output.image_url' | 'reasoning.encrypted_content' | 'code_interpreter_call.outputs' | 'web_search_call.action.return_token_budget'> | null;
     readonly instructions?: string | null;
     readonly max_output_tokens?: number | null;
     readonly metadata?: Readonly<Record<string, string>> | null;
     readonly parallel_tool_calls?: boolean | null;
     readonly previous_response_id?: string | null;
     readonly reasoning?: {
-        readonly effort?: 'minimal' | 'low' | 'medium' | 'high' | null;
+        readonly effort?: 'none' | 'minimal' | 'low' | 'medium' | 'high' | 'xhigh' | null;
         readonly summary?: 'auto' | 'concise' | 'detailed' | null;
     } | null;
     readonly service_tier?: 'auto' | 'default' | 'flex' | null;
@@ -7629,6 +7655,7 @@ interface AxAIOpenAIResponsesOutputMessageItem {
     role: 'assistant';
     content: ReadonlyArray<AxAIOpenAIResponsesOutputTextContentPart | AxAIOpenAIResponsesOutputRefusalContentPart>;
     status: 'in_progress' | 'completed' | 'incomplete';
+    phase?: 'commentary' | 'final_answer';
 }
 interface AxAIOpenAIResponsesFunctionCallItem {
     type: 'function_call';
@@ -7978,7 +8005,7 @@ type AxAIOpenAIResponsesConfig<TModel, TEmbedModel> = Omit<AxModelConfig, 'topK'
     logprobs?: number;
     echo?: boolean;
     dimensions?: number;
-    reasoningEffort?: 'minimal' | 'low' | 'medium' | 'high';
+    reasoningEffort?: 'none' | 'minimal' | 'low' | 'medium' | 'high' | 'xhigh';
     reasoningSummary?: 'auto' | 'concise' | 'detailed';
     store?: boolean;
     systemPrompt?: string;