@tyvm/knowhow 0.0.84 → 0.0.85

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@tyvm/knowhow",
-  "version": "0.0.84",
+  "version": "0.0.85",
   "description": "ai cli with plugins and agents",
   "main": "ts_build/src/index.js",
   "bin": {

package/src/agents/researcher/researcher.ts

@@ -8,7 +8,7 @@ export class ResearcherAgent extends BaseAgent {
 
   constructor(context: AgentContext) {
     super(context);
-    this.setModel(Models.google.Gemini_20_Flash);
+    this.setModel(Models.google.Gemini_3_Flash_Preview);
     this.setProvider("google");
     this.disableTool("patchFile");
     this.disableTool("writeFile");
@@ -106,4 +106,3 @@ export class ResearcherAgent extends BaseAgent {
     ] as Message[];
   }
 }
-

package/src/agents/tools/aiClient.ts

@@ -128,3 +128,51 @@ export async function listAllEmbeddingModels(): Promise<Record<string, string[]>
 
   return contextClients.listAllEmbeddingModels();
 }
+
+export async function listAllImageModels(): Promise<Record<string, string[]>> {
+  // Get context from bound ToolsService
+  const toolService = (
+    this instanceof ToolsService ? this : services().Tools
+  ) as ToolsService;
+
+  const toolContext = toolService.getContext();
+  const { Clients: contextClients } = toolContext;
+
+  if (!contextClients) {
+    throw new Error("Clients not available in tool context");
+  }
+
+  return contextClients.listAllImageModels();
+}
+
+export async function listAllAudioModels(): Promise<Record<string, string[]>> {
+  // Get context from bound ToolsService
+  const toolService = (
+    this instanceof ToolsService ? this : services().Tools
+  ) as ToolsService;
+
+  const toolContext = toolService.getContext();
+  const { Clients: contextClients } = toolContext;
+
+  if (!contextClients) {
+    throw new Error("Clients not available in tool context");
+  }
+
+  return contextClients.listAllAudioModels();
+}
+
+export async function listAllVideoModels(): Promise<Record<string, string[]>> {
+  // Get context from bound ToolsService
+  const toolService = (
+    this instanceof ToolsService ? this : services().Tools
+  ) as ToolsService;
+
+  const toolContext = toolService.getContext();
+  const { Clients: contextClients } = toolContext;
+
+  if (!contextClients) {
+    throw new Error("Clients not available in tool context");
+  }
+
+  return contextClients.listAllVideoModels();
+}

package/src/agents/tools/list.ts

@@ -506,6 +506,63 @@ export const includedTools = [
     },
   },
 
+  {
+    type: "function",
+    function: {
+      name: "listAllImageModels",
+      description:
+        "List all available image generation models using the knowhow ai client. Use this to discover which providers and models support image generation.",
+      parameters: {
+        type: "object",
+        properties: {},
+        required: [],
+      },
+      returns: {
+        type: "object",
+        description:
+          "A dictionary of all available image generation models for each provider",
+      },
+    },
+  },
+
+  {
+    type: "function",
+    function: {
+      name: "listAllAudioModels",
+      description:
+        "List all available audio generation models (TTS/transcription) using the knowhow ai client. Use this to discover which providers and models support audio generation.",
+      parameters: {
+        type: "object",
+        properties: {},
+        required: [],
+      },
+      returns: {
+        type: "object",
+        description:
+          "A dictionary of all available audio generation models for each provider",
+      },
+    },
+  },
+
+  {
+    type: "function",
+    function: {
+      name: "listAllVideoModels",
+      description:
+        "List all available video generation models using the knowhow ai client. Use this to discover which providers and models support video generation.",
+      parameters: {
+        type: "object",
+        properties: {},
+        required: [],
+      },
+      returns: {
+        type: "object",
+        description:
+          "A dictionary of all available video generation models for each provider",
+      },
+    },
+  },
+
   {
     type: "function",
     function: {

package/src/clients/index.ts

@@ -25,10 +25,23 @@ import { GenericGeminiClient } from "./gemini";
 import { HttpClient } from "./http";
 import { EmbeddingModels, Models } from "../types";
 import { getConfig } from "../config";
+import {
+  GoogleImageModels,
+  GoogleVideoModels,
+  GoogleTTSModels,
+  OpenAiImageModels,
+  OpenAiVideoModels,
+  OpenAiTTSModels,
+  OpenAiTranscriptionModels,
+  XaiImageModels,
+  XaiVideoModels,
+} from "../types";
 import { GenericXAIClient } from "./xai";
 import { KnowhowGenericClient } from "./knowhow";
 import { loadKnowhowJwt } from "../services/KnowhowClient";
 
+export type ModelModality = "completion" | "embedding" | "image" | "audio" | "video";
+
 function envCheck(key: string): boolean {
   const value = process.env[key];
   if (!value) {
@@ -85,6 +98,39 @@ export class AIClient {
     ...(envCheck("XAI_API_KEY") && { xai: this.completionModels.xai }),
   };
 
+  imageModels: Record<string, string[]> = {
+    ...(envCheck("OPENAI_KEY") && {
+      openai: OpenAiImageModels,
+    }),
+    ...(envCheck("GEMINI_API_KEY") && {
+      google: GoogleImageModels,
+    }),
+    ...(envCheck("XAI_API_KEY") && {
+      xai: XaiImageModels,
+    }),
+  };
+
+  audioModels: Record<string, string[]> = {
+    ...(envCheck("OPENAI_KEY") && {
+      openai: [...OpenAiTTSModels, ...OpenAiTranscriptionModels],
+    }),
+    ...(envCheck("GEMINI_API_KEY") && {
+      google: GoogleTTSModels,
+    }),
+  };
+
+  videoModels: Record<string, string[]> = {
+    ...(envCheck("OPENAI_KEY") && {
+      openai: OpenAiVideoModels,
+    }),
+    ...(envCheck("GEMINI_API_KEY") && {
+      google: GoogleVideoModels,
+    }),
+    ...(envCheck("XAI_API_KEY") && {
+      xai: XaiVideoModels,
+    }),
+  };
+
   getClient(provider: string, model?: string) {
     if (provider && !model) {
       return { client: this.clients[provider], provider, model: undefined };
@@ -215,6 +261,39 @@ export class AIClient {
     );
   }
 
+  registerImageModels(provider: string, models: string[]) {
+    const currentModels = this.clientModels[provider] || [];
+    const currentImageModels = this.imageModels[provider] || [];
+    this.clientModels[provider] = Array.from<string>(
+      new Set(currentModels.concat(models))
+    );
+    this.imageModels[provider] = Array.from<string>(
+      new Set(currentImageModels.concat(models))
+    );
+  }
+
+  registerAudioModels(provider: string, models: string[]) {
+    const currentModels = this.clientModels[provider] || [];
+    const currentAudioModels = this.audioModels[provider] || [];
+    this.clientModels[provider] = Array.from<string>(
+      new Set(currentModels.concat(models))
+    );
+    this.audioModels[provider] = Array.from<string>(
+      new Set(currentAudioModels.concat(models))
+    );
+  }
+
+  registerVideoModels(provider: string, models: string[]) {
+    const currentModels = this.clientModels[provider] || [];
+    const currentVideoModels = this.videoModels[provider] || [];
+    this.clientModels[provider] = Array.from<string>(
+      new Set(currentModels.concat(models))
+    );
+    this.videoModels[provider] = Array.from<string>(
+      new Set(currentVideoModels.concat(models))
+    );
+  }
+
   providerHasModel(provider: string, model: string): boolean {
     const models = this.clientModels[provider];
     if (!models) return false;
@@ -517,6 +596,18 @@ export class AIClient {
   listAllProviders() {
     return Object.keys(this.clientModels);
   }
+
+  listAllImageModels() {
+    return this.imageModels;
+  }
+
+  listAllAudioModels() {
+    return this.audioModels;
+  }
+
+  listAllVideoModels() {
+    return this.videoModels;
+  }
 }
 
 export const Clients = new AIClient();

package/src/clients/pricing/google.ts

@@ -1,7 +1,8 @@
 import { Models, EmbeddingModels } from "../../types";
 
 export const GeminiTextPricing = {
-  [Models.google.Gemini_3_Preview]: {
+  // Gemini 3.x
+  [Models.google.Gemini_31_Pro_Preview]: {
     input: 2,
     input_gt_200k: 4,
     output: 12,
@@ -9,6 +10,45 @@ export const GeminiTextPricing = {
     context_caching: 0.2,
     context_caching_gt_200k: 0.4,
   },
+  [Models.google.Gemini_31_Flash_Image_Preview]: {
+    input: 0.5,
+    output: 3,
+    image_generation: 0.045, // per 0.5K image
+  },
+  [Models.google.Gemini_31_Flash_Lite_Preview]: {
+    input: 0.25,
+    output: 1.5,
+    context_caching: 0.025,
+  },
+  [Models.google.Gemini_3_Flash_Preview]: {
+    input: 0.5,
+    output: 3.0,
+    context_caching: 0.05,
+  },
+  [Models.google.Gemini_3_Pro_Image_Preview]: {
+    input: 2,
+    output: 12,
+    image_generation: 0.134, // per 1K/2K image
+  },
+  // Gemini 2.5
+  [Models.google.Gemini_25_Pro]: {
+    input: 1.25,
+    input_gt_200k: 2.5,
+    output: 10.0,
+    output_gt_200k: 15.0,
+    context_caching: 0.125,
+    context_caching_gt_200k: 0.25,
+  },
+  [Models.google.Gemini_25_Flash]: {
+    input: 0.3,
+    output: 2.5,
+    context_caching: 0.03,
+  },
+  [Models.google.Gemini_25_Flash_Lite]: {
+    input: 0.1,
+    output: 0.4,
+    context_caching: 0.01,
+  },
   [Models.google.Gemini_25_Flash_Preview]: {
     input: 0.3,
     output: 2.5,
@@ -23,6 +63,19 @@ export const GeminiTextPricing = {
     context_caching: 0.125,
     context_caching_gt_200k: 0.25,
   },
+  [Models.google.Gemini_25_Flash_Image]: {
+    input: 0.3,
+    output: 0.039, // per image ($30/1M tokens, 1290 tokens per image)
+  },
+  [Models.google.Gemini_25_Flash_TTS]: {
+    input: 0.5,
+    output: 10.0,
+  },
+  [Models.google.Gemini_25_Pro_TTS]: {
+    input: 1.0,
+    output: 20.0,
+  },
+  // Gemini 2.0 (deprecated)
   [Models.google.Gemini_20_Flash]: {
     input: 0.1,
     output: 0.4,
@@ -37,6 +90,7 @@ export const GeminiTextPricing = {
     input: 0.075,
     output: 0.3,
   },
+  // Gemini 1.5 (legacy)
   [Models.google.Gemini_15_Flash]: {
     input: 0.075,
     output: 0.3,
@@ -52,14 +106,39 @@ export const GeminiTextPricing = {
     output: 5.0,
     context_caching: 0.3125,
   },
+  // Image generation
   [Models.google.Imagen_3]: {
-    image_generation: 0.03,
+    image_generation: 0.04, // Imagen 4 Standard: $0.04/image
+  },
+  [Models.google.Imagen_4_Fast]: {
+    image_generation: 0.02, // $0.02/image
   },
+  [Models.google.Imagen_4_Ultra]: {
+    image_generation: 0.06, // $0.06/image
+  },
+  // Video generation
   [Models.google.Veo_2]: {
     video_generation: 0.35,
   },
+  [Models.google.Veo_3]: {
+    video_generation: 0.4, // $0.40/second
+  },
+  [Models.google.Veo_3_Fast]: {
+    video_generation: 0.15, // $0.15/second
+  },
+  [Models.google.Veo_3_1]: {
+    video_generation: 0.4, // $0.40/second (720p/1080p)
+  },
+  [Models.google.Veo_3_1_Fast]: {
+    video_generation: 0.15, // $0.15/second
+  },
+  // Embeddings
   [EmbeddingModels.google.Gemini_Embedding]: {
     input: 0, // Free of charge
     output: 0, // Free of charge
   },
+  [EmbeddingModels.google.Gemini_Embedding_001]: {
+    input: 0.15,
+    output: 0,
+  },
 };

package/src/clients/pricing/openai.ts

@@ -1,6 +1,36 @@
 import { Models, EmbeddingModels } from "../../types";
 
 export const OpenAiTextPricing = {
+  [Models.openai.GPT_54]: {
+    input: 2.5,
+    cached_input: 0.25,
+    output: 15.0,
+  },
+  [Models.openai.GPT_54_Mini]: {
+    input: 0.75,
+    cached_input: 0.075,
+    output: 4.5,
+  },
+  [Models.openai.GPT_54_Nano]: {
+    input: 0.2,
+    cached_input: 0.02,
+    output: 1.25,
+  },
+  [Models.openai.GPT_54_Pro]: {
+    input: 30.0,
+    cached_input: 0,
+    output: 180.0,
+  },
+  [Models.openai.GPT_53_Chat]: {
+    input: 1.75,
+    cached_input: 0.175,
+    output: 14.0,
+  },
+  [Models.openai.GPT_53_Codex]: {
+    input: 1.75,
+    cached_input: 0.175,
+    output: 14.0,
+  },
   [Models.openai.GPT_4o]: {
     input: 2.5,
     cached_input: 1.25,
@@ -76,6 +106,11 @@ export const OpenAiTextPricing = {
     cached_input: 0.5,
     output: 8.0,
   },
+  [Models.openai.o3_Pro]: {
+    input: 20.0,
+    cached_input: 0,
+    output: 80.0,
+  },
   [Models.openai.o4_Mini]: {
     input: 1.1,
     cached_input: 0.275,
@@ -131,4 +166,37 @@ export const OpenAiTextPricing = {
     cached_input: 0,
     output: 0,
   },
+  // New realtime models
+  [Models.openai.GPT_Realtime_15]: {
+    input: 4.0,
+    cached_input: 0.4,
+    output: 16.0,
+  },
+  [Models.openai.GPT_Realtime_Mini]: {
+    input: 0.6,
+    cached_input: 0.06,
+    output: 2.4,
+  },
+  // New image models (text token pricing)
+  [Models.openai.GPT_Image_15]: {
+    input: 5.0,
+    cached_input: 1.25,
+    output: 10.0,
+  },
+  [Models.openai.GPT_Image_1_Mini]: {
+    input: 2.0,
+    cached_input: 0.2,
+    output: 0,
+  },
+  // New transcription models
+  [Models.openai.GPT_4o_Transcribe]: {
+    input: 2.5,
+    cached_input: 0,
+    output: 10.0,
+  },
+  [Models.openai.GPT_4o_Mini_Transcribe]: {
+    input: 1.25,
+    cached_input: 0,
+    output: 5.0,
+  },
 };

package/src/types.ts

@@ -216,6 +216,12 @@ export const Models = {
   openai: {
     GPT_5_2: "gpt-5.2",
     GPT_5_1: "gpt-5.1",
+    GPT_54: "gpt-5.4",
+    GPT_54_Mini: "gpt-5.4-mini",
+    GPT_54_Nano: "gpt-5.4-nano",
+    GPT_54_Pro: "gpt-5.4-pro",
+    GPT_53_Chat: "gpt-5.3-chat-latest",
+    GPT_53_Codex: "gpt-5.3-codex",
     GPT_5: "gpt-5",
     GPT_5_Mini: "gpt-5-mini",
     GPT_5_Nano: "gpt-5-nano",
@@ -239,6 +245,12 @@ export const Models = {
     GPT_4o_Mini_Search: "gpt-4o-mini-search-preview-2025-03-11",
     GPT_4o_Search: "gpt-4o-search-preview-2025-03-11",
 
+    GPT_4o_Transcribe: "gpt-4o-transcribe",
+    GPT_4o_Mini_Transcribe: "gpt-4o-mini-transcribe",
+    GPT_Realtime_15: "gpt-realtime-1.5",
+    GPT_Realtime_Mini: "gpt-realtime-mini",
+    GPT_Image_15: "gpt-image-1.5",
+    GPT_Image_1_Mini: "gpt-image-1-mini",
     TTS_1: "tts-1",
     Whisper_1: "whisper-1",
     DALL_E_3: "dall-e-3",
@@ -250,19 +262,41 @@ export const Models = {
     // Codex_Mini: "codex-mini-latest",
   },
   google: {
-    Gemini_3_Preview: "gemini-3-pro-preview",
+    // Gemini 3.x
+    Gemini_31_Pro_Preview: "gemini-3.1-pro-preview",
+    Gemini_31_Flash_Image_Preview: "gemini-3.1-flash-image-preview",
+    Gemini_31_Flash_Lite_Preview: "gemini-3.1-flash-lite-preview",
+    Gemini_3_Flash_Preview: "gemini-3-flash-preview",
+    Gemini_3_Pro_Image_Preview: "gemini-3-pro-image-preview",
+    // Gemini 2.5
+    Gemini_25_Pro: "gemini-2.5-pro",
+    Gemini_25_Flash: "gemini-2.5-flash",
+    Gemini_25_Flash_Lite: "gemini-2.5-flash-lite",
     Gemini_25_Flash_Preview: "gemini-2.5-flash-preview-05-20",
     Gemini_25_Pro_Preview: "gemini-2.5-pro-preview-05-06",
+    Gemini_25_Flash_Image: "gemini-2.5-flash-image",
+    Gemini_25_Flash_Live: "gemini-2.5-flash-live-preview",
+    Gemini_25_Flash_Native_Audio: "gemini-2.5-flash-native-audio-preview-12-2025",
+    Gemini_25_Pro_TTS: "gemini-2.5-pro-preview-tts",
+    // Gemini 2.0 (deprecated)
     Gemini_20_Flash: "gemini-2.0-flash",
     Gemini_20_Flash_Preview_Image_Generation:
       "gemini-2.0-flash-exp-image-generation",
     Gemini_20_Flash_Lite: "gemini-2.0-flash-lite",
+    // Gemini 1.5 (legacy)
     Gemini_15_Flash: "gemini-1.5-flash",
     Gemini_15_Flash_8B: "gemini-1.5-flash-8b",
     Gemini_15_Pro: "gemini-1.5-pro",
+    // Media generation
     Imagen_3: "imagen-4.0-generate-001",
+    Imagen_4_Fast: "imagen-4.0-fast-generate-001",
+    Imagen_4_Ultra: "imagen-4.0-ultra-generate-001",
     Veo_2: "veo-2.0-generate-001",
+    Veo_3: "veo-3.0-generate-001",
+    Veo_3_Fast: "veo-3.0-fast-generate-001",
     Veo_3_1: "veo-3.1-generate-preview",
+    Veo_3_1_Fast: "veo-3.1-fast-generate-preview",
+    // Audio / Live
     Gemini_20_Flash_Live: "gemini-2.0-flash-live-001",
     Gemini_25_Flash_TTS: "gemini-2.5-flash-preview-tts",
     Gemini_20_Flash_TTS: "gemini-2.0-flash-preview-tts",
@@ -277,6 +311,7 @@ export const EmbeddingModels = {
   },
   google: {
     Gemini_Embedding: "gemini-embedding-exp",
+    Gemini_Embedding_001: "gemini-embedding-001",
   },
 };
 
@@ -306,6 +341,12 @@ export const OpenAiReasoningModels = [
   Models.openai.o3,
   Models.openai.o3_Pro,
   Models.openai.o4_Mini,
+  Models.openai.GPT_54,
+  Models.openai.GPT_54_Mini,
+  Models.openai.GPT_54_Nano,
+  Models.openai.GPT_54_Pro,
+  Models.openai.GPT_53_Chat,
+  Models.openai.GPT_53_Codex,
   Models.openai.GPT_5,
   Models.openai.GPT_5_Mini,
   Models.openai.GPT_5_Nano,
@@ -321,6 +362,12 @@ export const OpenAiEmbeddingModels = [
 // export const OpenAiResponseOnlyModels = [Models.openai.Codex_Mini];
 
 export const GoogleReasoningModels = [
+  Models.google.Gemini_31_Pro_Preview,
+  Models.google.Gemini_31_Flash_Lite_Preview,
+  Models.google.Gemini_3_Flash_Preview,
+  Models.google.Gemini_25_Pro,
+  Models.google.Gemini_25_Flash,
+  Models.google.Gemini_25_Flash_Lite,
   Models.google.Gemini_25_Flash_Preview,
   Models.google.Gemini_25_Pro_Preview,
   Models.google.Gemini_20_Flash,
@@ -331,13 +378,20 @@ export const GoogleReasoningModels = [
 ];
 
 export const GoogleImageModels = [
+  Models.google.Gemini_31_Flash_Image_Preview,
+  Models.google.Gemini_3_Pro_Image_Preview,
+  Models.google.Gemini_25_Flash_Image,
   Models.google.Gemini_20_Flash_Preview_Image_Generation,
   Models.google.Imagen_3,
+  Models.google.Imagen_4_Fast,
+  Models.google.Imagen_4_Ultra,
 ];
 
 export const OpenAiImageModels = [
   Models.openai.DALL_E_3,
   Models.openai.DALL_E_2,
+  Models.openai.GPT_Image_15,
+  Models.openai.GPT_Image_1_Mini,
 ];
 
 export const OpenAiVideoModels = [
@@ -348,17 +402,36 @@ export const OpenAiVideoModels = [
 
 export const OpenAiTTSModels = [Models.openai.TTS_1];
 
-export const OpenAiTranscriptionModels = [Models.openai.Whisper_1];
-
 export const XaiImageModels = [Models.xai.GrokImagineImage];
+export const OpenAiTranscriptionModels = [
+  Models.openai.Whisper_1,
+  Models.openai.GPT_4o_Transcribe,
+  Models.openai.GPT_4o_Mini_Transcribe,
+];
 
+export const OpenAiRealtimeModels = [
+  Models.openai.GPT_4o_Realtime,
+  Models.openai.GPT_4o_Mini_Realtime,
+  Models.openai.GPT_Realtime_15,
+  Models.openai.GPT_Realtime_Mini,
+];
 export const XaiVideoModels = [Models.xai.GrokImagineVideo];
 
 export const GoogleTTSModels = [
   Models.google.Gemini_25_Flash_TTS,
+  Models.google.Gemini_25_Pro_TTS,
   Models.google.Gemini_20_Flash_TTS,
 ];
 
-export const GoogleVideoModels = [Models.google.Veo_2, Models.google.Veo_3_1];
+export const GoogleVideoModels = [
+  Models.google.Veo_2,
+  Models.google.Veo_3,
+  Models.google.Veo_3_Fast,
+  Models.google.Veo_3_1,
+  Models.google.Veo_3_1_Fast,
+];
 
-export const GoogleEmbeddingModels = [EmbeddingModels.google.Gemini_Embedding];
+export const GoogleEmbeddingModels = [
+  EmbeddingModels.google.Gemini_Embedding,
+  EmbeddingModels.google.Gemini_Embedding_001,
+];