@tanstack/ai-gemini 0.3.1 → 0.4.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -1 +1 @@
1
- {"version":3,"file":"model-meta.js","sources":["../../src/model-meta.ts"],"sourcesContent":["import type {\n GeminiCachedContentOptions,\n GeminiGenerationConfigOptions,\n GeminiSafetyOptions,\n GeminiStructuredOutputOptions,\n GeminiThinkingOptions,\n GeminiToolConfigOptions,\n} from './text/text-provider-options'\n\ninterface ModelMeta<TProviderOptions = unknown> {\n name: string\n supports: {\n input: Array<'text' | 'image' | 'audio' | 'video' | 'document'>\n output: Array<'text' | 'image' | 'audio' | 'video'>\n capabilities?: Array<\n | 'audio_generation'\n | 'batch_api'\n | 'caching'\n | 'code_execution'\n | 'file_search'\n | 'function_calling'\n | 'grounding_with_gmaps'\n | 'image_generation'\n | 'live_api'\n | 'search_grounding'\n | 'structured_output'\n | 'thinking'\n | 'url_context'\n >\n }\n max_input_tokens?: number\n max_output_tokens?: number\n knowledge_cutoff?: string\n pricing?: {\n input: {\n normal: number\n cached?: number\n }\n output: {\n normal: number\n }\n }\n /**\n * Type-level description of which provider options this model supports.\n */\n providerOptions?: TProviderOptions\n}\n\nconst GEMINI_3_PRO = {\n name: 'gemini-3-pro-preview',\n max_input_tokens: 1_048_576,\n max_output_tokens: 65_536,\n knowledge_cutoff: '2025-01-01',\n supports: {\n input: ['text', 'image', 'audio', 'video', 'document'],\n output: ['text'],\n capabilities: [\n 'batch_api',\n 'caching',\n 'code_execution',\n 'file_search',\n 'function_calling',\n 'search_grounding',\n 'structured_output',\n 'thinking',\n 'url_context',\n ],\n },\n pricing: {\n input: {\n normal: 2.5,\n },\n output: {\n normal: 15,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions\n>\n\nconst GEMINI_3_FLASH = {\n name: 'gemini-3-flash-preview',\n max_input_tokens: 1_048_576,\n max_output_tokens: 65_536,\n knowledge_cutoff: '2025-01-01',\n supports: {\n input: ['text', 'image', 'audio', 'video', 'document'],\n output: ['text'],\n capabilities: [\n 'batch_api',\n 'caching',\n 'code_execution',\n 'file_search',\n 'function_calling',\n 'search_grounding',\n 'structured_output',\n 'thinking',\n 'url_context',\n ],\n },\n pricing: {\n input: {\n normal: 0.5,\n },\n output: {\n normal: 3,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions\n>\n\nconst GEMINI_3_PRO_IMAGE = {\n name: 'gemini-3-pro-image-preview',\n max_input_tokens: 65_536,\n max_output_tokens: 32_768,\n knowledge_cutoff: '2025-01-01',\n supports: {\n input: ['text', 'image'],\n output: ['text', 'image'],\n capabilities: [\n 'batch_api',\n 'image_generation',\n 'search_grounding',\n 'structured_output',\n 'thinking',\n ],\n },\n pricing: {\n input: {\n normal: 2,\n },\n output: {\n normal: 0.134,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions\n>\n\nconst GEMINI_2_5_PRO = {\n name: 'gemini-2.5-pro',\n max_input_tokens: 1_048_576,\n max_output_tokens: 65_536,\n knowledge_cutoff: '2025-01-01',\n supports: {\n input: ['text', 'image', 'audio', 'video', 'document'],\n output: ['text'],\n capabilities: [\n 'batch_api',\n 'caching',\n 'code_execution',\n 'file_search',\n 'function_calling',\n 'grounding_with_gmaps',\n 'search_grounding',\n 'structured_output',\n 'thinking',\n 'url_context',\n ],\n },\n pricing: {\n input: {\n normal: 2.5,\n },\n output: {\n normal: 15,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions\n>\n\nconst GEMINI_2_5_PRO_TTS = {\n name: 'gemini-2.5-pro-preview-tts',\n max_input_tokens: 8_192,\n max_output_tokens: 16_384,\n knowledge_cutoff: '2025-05-01',\n supports: {\n input: ['text'],\n output: ['audio'],\n capabilities: ['audio_generation', 'file_search'],\n },\n pricing: {\n input: {\n normal: 2.5,\n },\n output: {\n normal: 15,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions\n>\n\nconst GEMINI_2_5_FLASH = {\n name: 'gemini-2.5-flash',\n max_input_tokens: 1_048_576,\n max_output_tokens: 65_536,\n knowledge_cutoff: '2025-01-01',\n supports: {\n input: ['text', 'image', 'audio', 'video'],\n output: ['text'],\n capabilities: [\n 'batch_api',\n 'caching',\n 'code_execution',\n 'file_search',\n 'function_calling',\n 'grounding_with_gmaps',\n 'search_grounding',\n 'structured_output',\n 'thinking',\n 'url_context',\n ],\n },\n pricing: {\n input: {\n normal: 1,\n },\n output: {\n normal: 2.5,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions\n>\n\nconst GEMINI_2_5_FLASH_PREVIEW = {\n name: 'gemini-2.5-flash-preview-09-2025',\n max_input_tokens: 1_048_576,\n max_output_tokens: 65_536,\n knowledge_cutoff: '2025-01-01',\n supports: {\n input: ['text', 'image', 'audio', 'video'],\n output: ['text'],\n capabilities: [\n 'batch_api',\n 'caching',\n 'code_execution',\n 'file_search',\n 'function_calling',\n 'search_grounding',\n 'structured_output',\n 'thinking',\n 'url_context',\n ],\n },\n pricing: {\n input: {\n normal: 1,\n },\n output: {\n normal: 2.5,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions\n>\n\nconst GEMINI_2_5_FLASH_IMAGE = {\n name: 'gemini-2.5-flash-image',\n max_input_tokens: 1_048_576,\n max_output_tokens: 65_536,\n knowledge_cutoff: '2025-06-01',\n supports: {\n input: ['text', 'image'],\n output: ['text', 'image'],\n capabilities: [\n 'batch_api',\n 'caching',\n 'file_search',\n 'image_generation',\n 'structured_output',\n ],\n },\n pricing: {\n input: {\n normal: 0.3,\n },\n output: {\n normal: 0.4,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions\n>\n/**\nconst GEMINI_2_5_FLASH_LIVE = {\n name: 'gemini-2.5-flash-native-audio-preview-09-2025',\n max_input_tokens: 141_072,\n max_output_tokens: 8_192,\n knowledge_cutoff: '2025-01-01',\n supports: {\n input: ['text', 'audio', 'video'],\n output: ['text', 'audio'],\n capabilities: [\n 'audio_generation',\n 'file_search',\n 'function_calling',\n 'live_api',\n 'search_grounding',\n 'thinking',\n ],\n },\n pricing: {\n // todo find this info\n input: {\n normal: 0,\n },\n output: {\n normal: 0,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions &\n GeminiThinkingOptions\n>\n*/\nconst GEMINI_2_5_FLASH_TTS = {\n name: 'gemini-2.5-flash-preview-tts',\n max_input_tokens: 8_192,\n max_output_tokens: 16_384,\n knowledge_cutoff: '2025-05-01',\n supports: {\n input: ['text'],\n output: ['audio'],\n capabilities: ['audio_generation', 'batch_api', 'file_search'],\n },\n pricing: {\n input: {\n normal: 1,\n },\n output: {\n normal: 2.5,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions\n>\n\nconst GEMINI_2_5_FLASH_LITE = {\n name: 'gemini-2.5-flash-lite',\n max_input_tokens: 1_048_576,\n max_output_tokens: 65_536,\n knowledge_cutoff: '2025-01-01',\n supports: {\n input: ['text', 'image', 'audio', 'video', 'document'],\n output: ['text'],\n capabilities: [\n 'batch_api',\n 'caching',\n 'code_execution',\n 'function_calling',\n 'grounding_with_gmaps',\n 'search_grounding',\n 'structured_output',\n 'thinking',\n 'url_context',\n ],\n },\n pricing: {\n input: {\n normal: 0.1,\n },\n output: {\n normal: 0.4,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions\n>\n\nconst GEMINI_2_5_FLASH_LITE_PREVIEW = {\n name: 'gemini-2.5-flash-lite-preview-09-2025',\n max_input_tokens: 1_048_576,\n max_output_tokens: 65_536,\n knowledge_cutoff: '2025-01-01',\n supports: {\n input: ['text', 'image', 'audio', 'video', 'document'],\n output: ['text'],\n capabilities: [\n 'batch_api',\n 'caching',\n 'code_execution',\n 'function_calling',\n 'search_grounding',\n 'structured_output',\n 'thinking',\n 'url_context',\n ],\n },\n pricing: {\n input: {\n normal: 0.1,\n },\n output: {\n normal: 0.4,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions\n>\n\nconst GEMINI_2_FLASH = {\n name: 'gemini-2.0-flash',\n max_input_tokens: 1_048_576,\n max_output_tokens: 8_192,\n knowledge_cutoff: '2024-08-01',\n supports: {\n input: ['text', 'image', 'audio', 'video'],\n output: ['text'],\n capabilities: [\n 'batch_api',\n 'caching',\n 'code_execution',\n 'function_calling',\n 'grounding_with_gmaps',\n 'live_api',\n 'search_grounding',\n 'structured_output',\n ],\n },\n pricing: {\n input: {\n normal: 0.1,\n },\n output: {\n normal: 0.4,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions\n>\n\nconst GEMINI_2_FLASH_IMAGE = {\n name: 'gemini-2.0-flash-preview-image-generation',\n max_input_tokens: 32_768,\n max_output_tokens: 8_192,\n knowledge_cutoff: '2024-08-01',\n supports: {\n input: ['text', 'image', 'audio', 'video'],\n output: ['text'],\n capabilities: [\n 'batch_api',\n 'caching',\n 'image_generation',\n 'structured_output',\n ],\n },\n pricing: {\n input: {\n normal: 0.1,\n },\n output: {\n normal: 0.039,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions\n>\n/* \nconst GEMINI_2_FLASH_LIVE = {\n name: 'gemini-2.0-flash-live-001',\n max_input_tokens: 1_048_576,\n max_output_tokens: 8_192,\n knowledge_cutoff: '2024-08-01',\n supports: {\n input: ['text', 'audio', 'video'],\n output: ['text', 'audio'],\n capabilities: [\n 'audio_generation',\n 'code_execution',\n 'function_calling',\n 'live_api',\n 'search_grounding',\n 'structured_output',\n 'url_context',\n ],\n },\n pricing: {\n // todo find this info\n input: {\n normal: 0,\n },\n output: {\n normal: 0,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions\n> */\n\nconst GEMINI_2_FLASH_LITE = {\n name: 'gemini-2.0-flash-lite',\n max_input_tokens: 1_048_576,\n max_output_tokens: 8_192,\n knowledge_cutoff: '2024-08-01',\n supports: {\n input: ['text', 'audio', 'video', 'image'],\n output: ['text'],\n capabilities: [\n 'batch_api',\n 'caching',\n 'function_calling',\n 'structured_output',\n ],\n },\n pricing: {\n input: {\n normal: 0.075,\n },\n output: {\n normal: 0.3,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions\n>\n\nconst IMAGEN_4_GENERATE = {\n name: 'imagen-4.0-generate-001',\n max_input_tokens: 480,\n max_output_tokens: 4,\n supports: {\n input: ['text'],\n output: ['image'],\n },\n pricing: {\n input: {\n normal: 0,\n },\n output: {\n normal: 0.4,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions\n>\n\nconst IMAGEN_4_GENERATE_ULTRA = {\n name: 'imagen-4.0-ultra-generate-001',\n max_input_tokens: 480,\n max_output_tokens: 4,\n supports: {\n input: ['text'],\n output: ['image'],\n },\n pricing: {\n input: {\n normal: 0,\n },\n output: {\n normal: 0.6,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions\n>\n\nconst IMAGEN_4_GENERATE_FAST = {\n name: 'imagen-4.0-fast-generate-001',\n max_input_tokens: 480,\n max_output_tokens: 4,\n supports: {\n input: ['text'],\n output: ['image'],\n },\n pricing: {\n input: {\n normal: 0,\n },\n output: {\n normal: 0.2,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions\n>\n\nconst IMAGEN_3 = {\n name: 'imagen-3.0-generate-002',\n max_output_tokens: 4,\n supports: {\n input: ['text'],\n output: ['image'],\n },\n pricing: {\n input: {\n normal: 0,\n },\n output: {\n normal: 0.03,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions\n>\n/** \nconst VEO_3_1_PREVIEW = {\n name: 'veo-3.1-generate-preview',\n max_input_tokens: 1024,\n max_output_tokens: 1,\n supports: {\n input: ['text', 'image'],\n output: ['video', 'audio'],\n },\n pricing: {\n input: {\n normal: 0,\n },\n output: {\n normal: 0.4,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions\n>\n\nconst VEO_3_1_FAST_PREVIEW = {\n name: 'veo-3.1-fast-generate-preview',\n max_input_tokens: 1024,\n max_output_tokens: 1,\n supports: {\n input: ['text', 'image'],\n output: ['video', 'audio'],\n },\n pricing: {\n input: {\n normal: 0,\n },\n output: {\n normal: 0.15,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions\n>\n\nconst VEO_3 = {\n name: 'veo-3.0-generate-001',\n max_input_tokens: 1024,\n max_output_tokens: 1,\n supports: {\n input: ['text', 'image'],\n output: ['video', 'audio'],\n },\n pricing: {\n input: {\n normal: 0,\n },\n output: {\n normal: 0.4,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions\n>\n\nconst VEO_3_FAST = {\n name: 'veo-3.0-fast-generate-001',\n max_input_tokens: 1024,\n max_output_tokens: 1,\n supports: {\n input: ['text', 'image'],\n output: ['video', 'audio'],\n },\n pricing: {\n input: {\n normal: 0,\n },\n output: {\n normal: 0.15,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions\n>\n\nconst VEO_2 = {\n name: 'veo-2.0-generate-001',\n max_output_tokens: 2,\n supports: {\n input: ['text', 'image'],\n output: ['video'],\n },\n pricing: {\n input: {\n normal: 0,\n },\n output: {\n normal: 0.35,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions\n> */\n\n/* const GEMINI_MODEL_META = {\n [GEMINI_3_PRO.name]: GEMINI_3_PRO,\n [GEMINI_2_5_PRO.name]: GEMINI_2_5_PRO,\n [GEMINI_2_5_PRO_TTS.name]: GEMINI_2_5_PRO_TTS,\n [GEMINI_2_5_FLASH.name]: GEMINI_2_5_FLASH,\n [GEMINI_2_5_FLASH_PREVIEW.name]: GEMINI_2_5_FLASH_PREVIEW,\n [GEMINI_2_5_FLASH_IMAGE.name]: GEMINI_2_5_FLASH_IMAGE,\n [GEMINI_2_5_FLASH_LIVE.name]: GEMINI_2_5_FLASH_LIVE,\n [GEMINI_2_5_FLASH_TTS.name]: GEMINI_2_5_FLASH_TTS,\n [GEMINI_2_5_FLASH_LITE.name]: GEMINI_2_5_FLASH_LITE,\n [GEMINI_2_5_FLASH_LITE_PREVIEW.name]: GEMINI_2_5_FLASH_LITE_PREVIEW,\n [GEMINI_2_FLASH.name]: GEMINI_2_FLASH,\n [GEMINI_2_FLASH_IMAGE.name]: GEMINI_2_FLASH_IMAGE,\n [GEMINI_2_FLASH_LIVE.name]: GEMINI_2_FLASH_LIVE,\n [GEMINI_2_FLASH_LITE.name]: GEMINI_2_FLASH_LITE,\n [IMAGEN_4_GENERATE.name]: IMAGEN_4_GENERATE,\n [IMAGEN_4_GENERATE_ULTRA.name]: IMAGEN_4_GENERATE_ULTRA,\n [IMAGEN_4_GENERATE_FAST.name]: IMAGEN_4_GENERATE_FAST,\n [IMAGEN_3.name]: IMAGEN_3,\n [VEO_3_1_PREVIEW.name]: VEO_3_1_PREVIEW,\n [VEO_3_1_FAST_PREVIEW.name]: VEO_3_1_FAST_PREVIEW,\n [VEO_3.name]: VEO_3,\n [VEO_3_FAST.name]: VEO_3_FAST,\n [VEO_2.name]: VEO_2,\n} as const */\n\nexport const GEMINI_MODELS = [\n GEMINI_3_PRO.name,\n GEMINI_3_FLASH.name,\n GEMINI_2_5_PRO.name,\n GEMINI_2_5_FLASH.name,\n GEMINI_2_5_FLASH_PREVIEW.name,\n GEMINI_2_5_FLASH_LITE.name,\n GEMINI_2_5_FLASH_LITE_PREVIEW.name,\n GEMINI_2_FLASH.name,\n GEMINI_2_FLASH_LITE.name,\n] as const\n\nexport type GeminiModels = (typeof GEMINI_MODELS)[number]\n\nexport type GeminiImageModels = (typeof GEMINI_IMAGE_MODELS)[number]\n\nexport const GEMINI_IMAGE_MODELS = [\n GEMINI_3_PRO_IMAGE.name,\n GEMINI_2_5_FLASH_IMAGE.name,\n GEMINI_2_FLASH_IMAGE.name,\n IMAGEN_3.name,\n IMAGEN_4_GENERATE.name,\n IMAGEN_4_GENERATE_FAST.name,\n IMAGEN_4_GENERATE_ULTRA.name,\n] as const\n\n/**\n * Text-to-speech models\n * @experimental Gemini TTS is an experimental feature and may change.\n */\nexport const GEMINI_TTS_MODELS = [\n GEMINI_2_5_FLASH_TTS.name,\n GEMINI_2_5_PRO_TTS.name,\n] as const\n\n/**\n * Available voice names for Gemini TTS\n * @see https://ai.google.dev/gemini-api/docs/speech-generation\n */\nexport const GEMINI_TTS_VOICES = [\n 'Zephyr',\n 'Puck',\n 'Charon',\n 'Kore',\n 'Fenrir',\n 'Leda',\n 'Orus',\n 'Aoede',\n 'Callirrhoe',\n 'Autonoe',\n 'Enceladus',\n 'Iapetus',\n 'Umbriel',\n 'Algieba',\n 'Despina',\n 'Erinome',\n 'Algenib',\n 'Rasalgethi',\n 'Laomedeia',\n 'Achernar',\n 'Alnilam',\n 'Schedar',\n 'Gacrux',\n 'Pulcherrima',\n 'Achird',\n 'Zubenelgenubi',\n 'Vindemiatrix',\n 'Sadachbia',\n 'Sadaltager',\n 'Sulafat',\n] as const\n\nexport type GeminiTTSVoice = (typeof GEMINI_TTS_VOICES)[number]\n\n/* const GEMINI_AUDIO_MODELS = [\n GEMINI_2_5_PRO_TTS.name,\n GEMINI_2_5_FLASH_TTS.name,\n GEMINI_2_5_FLASH_LIVE.name,\n GEMINI_2_FLASH_LIVE.name,\n] as const\n\n const GEMINI_VIDEO_MODELS = [\n VEO_3_1_PREVIEW.name,\n VEO_3_1_FAST_PREVIEW.name,\n VEO_3.name,\n VEO_3_FAST.name,\n VEO_2.name,\n] as const */\n\n// Manual type map for per-model provider options\nexport type GeminiChatModelProviderOptionsByName = {\n // Models with thinking and structured output support\n [GEMINI_3_PRO.name]: GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions\n [GEMINI_3_FLASH.name]: GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions\n [GEMINI_2_5_PRO.name]: GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions\n [GEMINI_2_5_FLASH.name]: GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions\n [GEMINI_2_5_FLASH_PREVIEW.name]: GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions\n [GEMINI_2_5_FLASH_LITE.name]: GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions\n [GEMINI_2_5_FLASH_LITE_PREVIEW.name]: GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions\n // Models with structured output but no thinking support\n [GEMINI_2_FLASH.name]: GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions\n [GEMINI_2_FLASH_LITE.name]: GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions\n}\n\n/**\n * Type-only map from chat model name to its supported input modalities.\n * Based on the 'supports.input' arrays defined for each model.\n * Note: 'document' in the model meta is mapped to 'document' modality.\n * Used by the core AI types to constrain ContentPart types based on the selected model.\n * Note: These must be inlined as readonly arrays (not typeof) because the model\n * constants are not exported and typeof references don't work in .d.ts files\n * when consumed by external packages.\n *\n * @see https://ai.google.dev/gemini-api/docs/vision\n * @see https://ai.google.dev/gemini-api/docs/audio\n * @see https://ai.google.dev/gemini-api/docs/document-processing\n */\nexport type GeminiModelInputModalitiesByName = {\n // Models with full multimodal support (text, image, audio, video, document)\n [GEMINI_3_PRO.name]: typeof GEMINI_3_PRO.supports.input\n [GEMINI_3_FLASH.name]: typeof GEMINI_3_FLASH.supports.input\n [GEMINI_2_5_PRO.name]: typeof GEMINI_2_5_PRO.supports.input\n [GEMINI_2_5_FLASH_LITE.name]: typeof GEMINI_2_5_FLASH_LITE.supports.input\n [GEMINI_2_5_FLASH_LITE_PREVIEW.name]: typeof GEMINI_2_5_FLASH_LITE_PREVIEW.supports.input\n\n // Models with text, image, audio, video (no document)\n [GEMINI_2_5_FLASH.name]: typeof GEMINI_2_5_FLASH.supports.input\n [GEMINI_2_5_FLASH_PREVIEW.name]: typeof GEMINI_2_5_FLASH_PREVIEW.supports.input\n [GEMINI_2_FLASH.name]: typeof GEMINI_2_FLASH.supports.input\n [GEMINI_2_FLASH_LITE.name]: typeof GEMINI_2_FLASH_LITE.supports.input\n}\n"],"names":[],"mappings":"AAgDA,MAAM,eAAe;AAAA,EACnB,MAAM;AA2BR;AASA,MAAM,iBAAiB;AAAA,EACrB,MAAM;AA2BR;AASA,MAAM,qBAAqB;AAAA,EACzB,MAAM;AAuBR;AASA,MAAM,iBAAiB;AAAA,EACrB,MAAM;AA4BR;AASA,MAAM,qBAAqB;AAAA,EACzB,MAAM;AAiBR;AAOA,MAAM,mBAAmB;AAAA,EACvB,MAAM;AA4BR;AASA,MAAM,2BAA2B;AAAA,EAC/B,MAAM;AA2BR;AASA,MAAM,yBAAyB;AAAA,EAC7B,MAAM;AAuBR;AAyCA,MAAM,uBAAuB;AAAA,EAC3B,MAAM;AAiBR;AAOA,MAAM,wBAAwB;AAAA,EAC5B,MAAM;AA2BR;AASA,MAAM,gCAAgC;AAAA,EACpC,MAAM;AA0BR;AASA,MAAM,iBAAiB;AAAA,EACrB,MAAM;AA0BR;AAQA,MAAM,uBAAuB;AAAA,EAC3B,MAAM;AAsBR;AAyCA,MAAM,sBAAsB;AAAA,EAC1B,MAAM;AAsBR;AAQA,MAAM,oBAAoB;AAAA,EACxB,MAAM;AAeR;AAOA,MAAM,0BAA0B;AAAA,EAC9B,MAAM;AAeR;AAOA,MAAM,yBAAyB;AAAA,EAC7B,MAAM;AAeR;AAOA,MAAM,WAAW;AAAA,EACf,MAAM;AAcR;AAmJO,MAAM,gBAAgB;AAAA,EAC3B,aAAa;AAAA,EACb,eAAe;AAAA,EACf,eAAe;AAAA,EACf,iBAAiB;AAAA,EACjB,yBAAyB;AAAA,EACzB,sBAAsB;AAAA,EACtB,8BAA8B;AAAA,EAC9B,eAAe;AAAA,EACf,oBAAoB;AACtB;AAMO,MAAM,sBAAsB;AAAA,EACjC,mBAAmB;AAAA,EACnB,uBAAuB;AAAA,EACvB,qBAAqB;AAAA,EACrB,SAAS;AAAA,EACT,kBAAkB;AAAA,EAClB,uBAAuB;AAAA,EACvB,wBAAwB;AAC1B;AAMO,MAAM,oBAAoB;AAAA,EAC/B,qBAAqB;AAAA,EACrB,mBAAmB;AACrB;AAMO,MAAM,oBAAoB;AAAA,EAC/B;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AACF;"}
1
+ {"version":3,"file":"model-meta.js","sources":["../../src/model-meta.ts"],"sourcesContent":["import type {\n GeminiCachedContentOptions,\n GeminiCommonConfigOptions,\n GeminiSafetyOptions,\n GeminiStructuredOutputOptions,\n GeminiThinkingAdvancedOptions,\n GeminiThinkingOptions,\n GeminiToolConfigOptions,\n} from './text/text-provider-options'\n\ninterface ModelMeta<TProviderOptions = unknown> {\n name: string\n supports: {\n input: Array<'text' | 'image' | 'audio' | 'video' | 'document'>\n output: Array<'text' | 'image' | 'audio' | 'video'>\n capabilities?: Array<\n | 'audio_generation'\n | 'batch_api'\n | 'caching'\n | 'code_execution'\n | 'file_search'\n | 'function_calling'\n | 'grounding_with_gmaps'\n | 'image_generation'\n | 'live_api'\n | 'search_grounding'\n | 'structured_output'\n | 'thinking'\n | 'url_context'\n >\n }\n max_input_tokens?: number\n max_output_tokens?: number\n knowledge_cutoff?: string\n pricing?: {\n input: {\n normal: number\n cached?: number\n }\n output: {\n normal: number\n }\n }\n /**\n * Type-level description of which provider options this model supports.\n */\n providerOptions?: TProviderOptions\n}\n\nconst GEMINI_3_PRO = {\n name: 'gemini-3-pro-preview',\n max_input_tokens: 1_048_576,\n max_output_tokens: 65_536,\n knowledge_cutoff: '2025-01-01',\n supports: {\n input: ['text', 'image', 'audio', 'video', 'document'],\n output: ['text'],\n capabilities: [\n 'batch_api',\n 'caching',\n 'code_execution',\n 'file_search',\n 'function_calling',\n 'search_grounding',\n 'structured_output',\n 'thinking',\n 'url_context',\n ],\n },\n pricing: {\n input: {\n normal: 2.5,\n },\n output: {\n normal: 15,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiCommonConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions &\n GeminiThinkingAdvancedOptions\n>\n\nconst GEMINI_3_FLASH = {\n name: 'gemini-3-flash-preview',\n max_input_tokens: 1_048_576,\n max_output_tokens: 65_536,\n knowledge_cutoff: '2025-01-01',\n supports: {\n input: ['text', 'image', 'audio', 'video', 'document'],\n output: ['text'],\n capabilities: [\n 'batch_api',\n 'caching',\n 'code_execution',\n 'file_search',\n 'function_calling',\n 'search_grounding',\n 'structured_output',\n 'thinking',\n 'url_context',\n ],\n },\n pricing: {\n input: {\n normal: 0.5,\n },\n output: {\n normal: 3,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiCommonConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions &\n GeminiThinkingAdvancedOptions\n>\n\nconst GEMINI_3_PRO_IMAGE = {\n name: 'gemini-3-pro-image-preview',\n max_input_tokens: 65_536,\n max_output_tokens: 32_768,\n knowledge_cutoff: '2025-01-01',\n supports: {\n input: ['text', 'image'],\n output: ['text', 'image'],\n capabilities: [\n 'batch_api',\n 'image_generation',\n 'search_grounding',\n 'structured_output',\n 'thinking',\n ],\n },\n pricing: {\n input: {\n normal: 2,\n },\n output: {\n normal: 0.134,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiCommonConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions &\n GeminiThinkingAdvancedOptions\n>\n\nconst GEMINI_2_5_PRO = {\n name: 'gemini-2.5-pro',\n max_input_tokens: 1_048_576,\n max_output_tokens: 65_536,\n knowledge_cutoff: '2025-01-01',\n supports: {\n input: ['text', 'image', 'audio', 'video', 'document'],\n output: ['text'],\n capabilities: [\n 'batch_api',\n 'caching',\n 'code_execution',\n 'file_search',\n 'function_calling',\n 'grounding_with_gmaps',\n 'search_grounding',\n 'structured_output',\n 'thinking',\n 'url_context',\n ],\n },\n pricing: {\n input: {\n normal: 2.5,\n },\n output: {\n normal: 15,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiCommonConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions\n>\n\nconst GEMINI_2_5_PRO_TTS = {\n name: 'gemini-2.5-pro-preview-tts',\n max_input_tokens: 8_192,\n max_output_tokens: 16_384,\n knowledge_cutoff: '2025-05-01',\n supports: {\n input: ['text'],\n output: ['audio'],\n capabilities: ['audio_generation', 'file_search'],\n },\n pricing: {\n input: {\n normal: 2.5,\n },\n output: {\n normal: 15,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiCommonConfigOptions &\n GeminiCachedContentOptions\n>\n\nconst GEMINI_2_5_FLASH = {\n name: 'gemini-2.5-flash',\n max_input_tokens: 1_048_576,\n max_output_tokens: 65_536,\n knowledge_cutoff: '2025-01-01',\n supports: {\n input: ['text', 'image', 'audio', 'video'],\n output: ['text'],\n capabilities: [\n 'batch_api',\n 'caching',\n 'code_execution',\n 'file_search',\n 'function_calling',\n 'grounding_with_gmaps',\n 'search_grounding',\n 'structured_output',\n 'thinking',\n 'url_context',\n ],\n },\n pricing: {\n input: {\n normal: 1,\n },\n output: {\n normal: 2.5,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiCommonConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions\n>\n\nconst GEMINI_2_5_FLASH_PREVIEW = {\n name: 'gemini-2.5-flash-preview-09-2025',\n max_input_tokens: 1_048_576,\n max_output_tokens: 65_536,\n knowledge_cutoff: '2025-01-01',\n supports: {\n input: ['text', 'image', 'audio', 'video'],\n output: ['text'],\n capabilities: [\n 'batch_api',\n 'caching',\n 'code_execution',\n 'file_search',\n 'function_calling',\n 'search_grounding',\n 'structured_output',\n 'thinking',\n 'url_context',\n ],\n },\n pricing: {\n input: {\n normal: 1,\n },\n output: {\n normal: 2.5,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiCommonConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions\n>\n\nconst GEMINI_2_5_FLASH_IMAGE = {\n name: 'gemini-2.5-flash-image',\n max_input_tokens: 1_048_576,\n max_output_tokens: 65_536,\n knowledge_cutoff: '2025-06-01',\n supports: {\n input: ['text', 'image'],\n output: ['text', 'image'],\n capabilities: [\n 'batch_api',\n 'caching',\n 'file_search',\n 'image_generation',\n 'structured_output',\n ],\n },\n pricing: {\n input: {\n normal: 0.3,\n },\n output: {\n normal: 0.4,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiCommonConfigOptions &\n GeminiCachedContentOptions\n>\n/**\nconst GEMINI_2_5_FLASH_LIVE = {\n name: 'gemini-2.5-flash-native-audio-preview-09-2025',\n max_input_tokens: 141_072,\n max_output_tokens: 8_192,\n knowledge_cutoff: '2025-01-01',\n supports: {\n input: ['text', 'audio', 'video'],\n output: ['text', 'audio'],\n capabilities: [\n 'audio_generation',\n 'file_search',\n 'function_calling',\n 'live_api',\n 'search_grounding',\n 'thinking',\n ],\n },\n pricing: {\n // todo find this info\n input: {\n normal: 0,\n },\n output: {\n normal: 0,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions &\n GeminiThinkingOptions\n>\n*/\nconst GEMINI_2_5_FLASH_TTS = {\n name: 'gemini-2.5-flash-preview-tts',\n max_input_tokens: 8_192,\n max_output_tokens: 16_384,\n knowledge_cutoff: '2025-05-01',\n supports: {\n input: ['text'],\n output: ['audio'],\n capabilities: ['audio_generation', 'batch_api', 'file_search'],\n },\n pricing: {\n input: {\n normal: 1,\n },\n output: {\n normal: 2.5,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiCommonConfigOptions &\n GeminiCachedContentOptions\n>\n\nconst GEMINI_2_5_FLASH_LITE = {\n name: 'gemini-2.5-flash-lite',\n max_input_tokens: 1_048_576,\n max_output_tokens: 65_536,\n knowledge_cutoff: '2025-01-01',\n supports: {\n input: ['text', 'image', 'audio', 'video', 'document'],\n output: ['text'],\n capabilities: [\n 'batch_api',\n 'caching',\n 'code_execution',\n 'function_calling',\n 'grounding_with_gmaps',\n 'search_grounding',\n 'structured_output',\n 'thinking',\n 'url_context',\n ],\n },\n pricing: {\n input: {\n normal: 0.1,\n },\n output: {\n normal: 0.4,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiCommonConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions\n>\n\nconst GEMINI_2_5_FLASH_LITE_PREVIEW = {\n name: 'gemini-2.5-flash-lite-preview-09-2025',\n max_input_tokens: 1_048_576,\n max_output_tokens: 65_536,\n knowledge_cutoff: '2025-01-01',\n supports: {\n input: ['text', 'image', 'audio', 'video', 'document'],\n output: ['text'],\n capabilities: [\n 'batch_api',\n 'caching',\n 'code_execution',\n 'function_calling',\n 'search_grounding',\n 'structured_output',\n 'thinking',\n 'url_context',\n ],\n },\n pricing: {\n input: {\n normal: 0.1,\n },\n output: {\n normal: 0.4,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiCommonConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions\n>\n\nconst GEMINI_2_FLASH = {\n name: 'gemini-2.0-flash',\n max_input_tokens: 1_048_576,\n max_output_tokens: 8_192,\n knowledge_cutoff: '2024-08-01',\n supports: {\n input: ['text', 'image', 'audio', 'video'],\n output: ['text'],\n capabilities: [\n 'batch_api',\n 'caching',\n 'code_execution',\n 'function_calling',\n 'grounding_with_gmaps',\n 'live_api',\n 'search_grounding',\n 'structured_output',\n ],\n },\n pricing: {\n input: {\n normal: 0.1,\n },\n output: {\n normal: 0.4,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiCommonConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions\n>\n\nconst GEMINI_2_FLASH_IMAGE = {\n name: 'gemini-2.0-flash-preview-image-generation',\n max_input_tokens: 32_768,\n max_output_tokens: 8_192,\n knowledge_cutoff: '2024-08-01',\n supports: {\n input: ['text', 'image', 'audio', 'video'],\n output: ['text'],\n capabilities: [\n 'batch_api',\n 'caching',\n 'image_generation',\n 'structured_output',\n ],\n },\n pricing: {\n input: {\n normal: 0.1,\n },\n output: {\n normal: 0.039,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiCommonConfigOptions &\n GeminiCachedContentOptions\n>\n/* \nconst GEMINI_2_FLASH_LIVE = {\n name: 'gemini-2.0-flash-live-001',\n max_input_tokens: 1_048_576,\n max_output_tokens: 8_192,\n knowledge_cutoff: '2024-08-01',\n supports: {\n input: ['text', 'audio', 'video'],\n output: ['text', 'audio'],\n capabilities: [\n 'audio_generation',\n 'code_execution',\n 'function_calling',\n 'live_api',\n 'search_grounding',\n 'structured_output',\n 'url_context',\n ],\n },\n pricing: {\n // todo find this info\n input: {\n normal: 0,\n },\n output: {\n normal: 0,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions\n> */\n\nconst GEMINI_2_FLASH_LITE = {\n name: 'gemini-2.0-flash-lite',\n max_input_tokens: 1_048_576,\n max_output_tokens: 8_192,\n knowledge_cutoff: '2024-08-01',\n supports: {\n input: ['text', 'audio', 'video', 'image'],\n output: ['text'],\n capabilities: [\n 'batch_api',\n 'caching',\n 'function_calling',\n 'structured_output',\n ],\n },\n pricing: {\n input: {\n normal: 0.075,\n },\n output: {\n normal: 0.3,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiCommonConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions\n>\n\nconst IMAGEN_4_GENERATE = {\n name: 'imagen-4.0-generate-001',\n max_input_tokens: 480,\n max_output_tokens: 4,\n supports: {\n input: ['text'],\n output: ['image'],\n },\n pricing: {\n input: {\n normal: 0,\n },\n output: {\n normal: 0.4,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiCommonConfigOptions &\n GeminiCachedContentOptions\n>\n\nconst IMAGEN_4_GENERATE_ULTRA = {\n name: 'imagen-4.0-ultra-generate-001',\n max_input_tokens: 480,\n max_output_tokens: 4,\n supports: {\n input: ['text'],\n output: ['image'],\n },\n pricing: {\n input: {\n normal: 0,\n },\n output: {\n normal: 0.6,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiCommonConfigOptions &\n GeminiCachedContentOptions\n>\n\nconst IMAGEN_4_GENERATE_FAST = {\n name: 'imagen-4.0-fast-generate-001',\n max_input_tokens: 480,\n max_output_tokens: 4,\n supports: {\n input: ['text'],\n output: ['image'],\n },\n pricing: {\n input: {\n normal: 0,\n },\n output: {\n normal: 0.2,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiCommonConfigOptions &\n GeminiCachedContentOptions\n>\n\nconst IMAGEN_3 = {\n name: 'imagen-3.0-generate-002',\n max_output_tokens: 4,\n supports: {\n input: ['text'],\n output: ['image'],\n },\n pricing: {\n input: {\n normal: 0,\n },\n output: {\n normal: 0.03,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiCommonConfigOptions &\n GeminiCachedContentOptions\n>\n/** \nconst VEO_3_1_PREVIEW = {\n name: 'veo-3.1-generate-preview',\n max_input_tokens: 1024,\n max_output_tokens: 1,\n supports: {\n input: ['text', 'image'],\n output: ['video', 'audio'],\n },\n pricing: {\n input: {\n normal: 0,\n },\n output: {\n normal: 0.4,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions\n>\n\nconst VEO_3_1_FAST_PREVIEW = {\n name: 'veo-3.1-fast-generate-preview',\n max_input_tokens: 1024,\n max_output_tokens: 1,\n supports: {\n input: ['text', 'image'],\n output: ['video', 'audio'],\n },\n pricing: {\n input: {\n normal: 0,\n },\n output: {\n normal: 0.15,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions\n>\n\nconst VEO_3 = {\n name: 'veo-3.0-generate-001',\n max_input_tokens: 1024,\n max_output_tokens: 1,\n supports: {\n input: ['text', 'image'],\n output: ['video', 'audio'],\n },\n pricing: {\n input: {\n normal: 0,\n },\n output: {\n normal: 0.4,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions\n>\n\nconst VEO_3_FAST = {\n name: 'veo-3.0-fast-generate-001',\n max_input_tokens: 1024,\n max_output_tokens: 1,\n supports: {\n input: ['text', 'image'],\n output: ['video', 'audio'],\n },\n pricing: {\n input: {\n normal: 0,\n },\n output: {\n normal: 0.15,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions\n>\n\nconst VEO_2 = {\n name: 'veo-2.0-generate-001',\n max_output_tokens: 2,\n supports: {\n input: ['text', 'image'],\n output: ['video'],\n },\n pricing: {\n input: {\n normal: 0,\n },\n output: {\n normal: 0.35,\n },\n },\n} as const satisfies ModelMeta<\n GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiGenerationConfigOptions &\n GeminiCachedContentOptions\n> */\n\n/* const GEMINI_MODEL_META = {\n [GEMINI_3_PRO.name]: GEMINI_3_PRO,\n [GEMINI_2_5_PRO.name]: GEMINI_2_5_PRO,\n [GEMINI_2_5_PRO_TTS.name]: GEMINI_2_5_PRO_TTS,\n [GEMINI_2_5_FLASH.name]: GEMINI_2_5_FLASH,\n [GEMINI_2_5_FLASH_PREVIEW.name]: GEMINI_2_5_FLASH_PREVIEW,\n [GEMINI_2_5_FLASH_IMAGE.name]: GEMINI_2_5_FLASH_IMAGE,\n [GEMINI_2_5_FLASH_LIVE.name]: GEMINI_2_5_FLASH_LIVE,\n [GEMINI_2_5_FLASH_TTS.name]: GEMINI_2_5_FLASH_TTS,\n [GEMINI_2_5_FLASH_LITE.name]: GEMINI_2_5_FLASH_LITE,\n [GEMINI_2_5_FLASH_LITE_PREVIEW.name]: GEMINI_2_5_FLASH_LITE_PREVIEW,\n [GEMINI_2_FLASH.name]: GEMINI_2_FLASH,\n [GEMINI_2_FLASH_IMAGE.name]: GEMINI_2_FLASH_IMAGE,\n [GEMINI_2_FLASH_LIVE.name]: GEMINI_2_FLASH_LIVE,\n [GEMINI_2_FLASH_LITE.name]: GEMINI_2_FLASH_LITE,\n [IMAGEN_4_GENERATE.name]: IMAGEN_4_GENERATE,\n [IMAGEN_4_GENERATE_ULTRA.name]: IMAGEN_4_GENERATE_ULTRA,\n [IMAGEN_4_GENERATE_FAST.name]: IMAGEN_4_GENERATE_FAST,\n [IMAGEN_3.name]: IMAGEN_3,\n [VEO_3_1_PREVIEW.name]: VEO_3_1_PREVIEW,\n [VEO_3_1_FAST_PREVIEW.name]: VEO_3_1_FAST_PREVIEW,\n [VEO_3.name]: VEO_3,\n [VEO_3_FAST.name]: VEO_3_FAST,\n [VEO_2.name]: VEO_2,\n} as const */\n\nexport const GEMINI_MODELS = [\n GEMINI_3_PRO.name,\n GEMINI_3_FLASH.name,\n GEMINI_2_5_PRO.name,\n GEMINI_2_5_FLASH.name,\n GEMINI_2_5_FLASH_PREVIEW.name,\n GEMINI_2_5_FLASH_LITE.name,\n GEMINI_2_5_FLASH_LITE_PREVIEW.name,\n GEMINI_2_FLASH.name,\n GEMINI_2_FLASH_LITE.name,\n] as const\n\nexport type GeminiModels = (typeof GEMINI_MODELS)[number]\n\nexport type GeminiImageModels = (typeof GEMINI_IMAGE_MODELS)[number]\n\nexport const GEMINI_IMAGE_MODELS = [\n GEMINI_3_PRO_IMAGE.name,\n GEMINI_2_5_FLASH_IMAGE.name,\n GEMINI_2_FLASH_IMAGE.name,\n IMAGEN_3.name,\n IMAGEN_4_GENERATE.name,\n IMAGEN_4_GENERATE_FAST.name,\n IMAGEN_4_GENERATE_ULTRA.name,\n] as const\n\n/**\n * Text-to-speech models\n * @experimental Gemini TTS is an experimental feature and may change.\n */\nexport const GEMINI_TTS_MODELS = [\n GEMINI_2_5_FLASH_TTS.name,\n GEMINI_2_5_PRO_TTS.name,\n] as const\n\n/**\n * Available voice names for Gemini TTS\n * @see https://ai.google.dev/gemini-api/docs/speech-generation\n */\nexport const GEMINI_TTS_VOICES = [\n 'Zephyr',\n 'Puck',\n 'Charon',\n 'Kore',\n 'Fenrir',\n 'Leda',\n 'Orus',\n 'Aoede',\n 'Callirrhoe',\n 'Autonoe',\n 'Enceladus',\n 'Iapetus',\n 'Umbriel',\n 'Algieba',\n 'Despina',\n 'Erinome',\n 'Algenib',\n 'Rasalgethi',\n 'Laomedeia',\n 'Achernar',\n 'Alnilam',\n 'Schedar',\n 'Gacrux',\n 'Pulcherrima',\n 'Achird',\n 'Zubenelgenubi',\n 'Vindemiatrix',\n 'Sadachbia',\n 'Sadaltager',\n 'Sulafat',\n] as const\n\nexport type GeminiTTSVoice = (typeof GEMINI_TTS_VOICES)[number]\n\n/* const GEMINI_AUDIO_MODELS = [\n GEMINI_2_5_PRO_TTS.name,\n GEMINI_2_5_FLASH_TTS.name,\n GEMINI_2_5_FLASH_LIVE.name,\n GEMINI_2_FLASH_LIVE.name,\n] as const\n\n const GEMINI_VIDEO_MODELS = [\n VEO_3_1_PREVIEW.name,\n VEO_3_1_FAST_PREVIEW.name,\n VEO_3.name,\n VEO_3_FAST.name,\n VEO_2.name,\n] as const */\n\n// Manual type map for per-model provider options\nexport type GeminiChatModelProviderOptionsByName = {\n // Models with thinking and structured output support\n [GEMINI_3_PRO.name]: GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiCommonConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions &\n GeminiThinkingAdvancedOptions\n [GEMINI_3_FLASH.name]: GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiCommonConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions &\n GeminiThinkingAdvancedOptions\n [GEMINI_2_5_PRO.name]: GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiCommonConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions\n [GEMINI_2_5_FLASH.name]: GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiCommonConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions\n [GEMINI_2_5_FLASH_PREVIEW.name]: GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiCommonConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions\n [GEMINI_2_5_FLASH_LITE.name]: GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiCommonConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions\n [GEMINI_2_5_FLASH_LITE_PREVIEW.name]: GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiCommonConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions &\n GeminiThinkingOptions\n // Models with structured output but no thinking support\n [GEMINI_2_FLASH.name]: GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiCommonConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions\n [GEMINI_2_FLASH_LITE.name]: GeminiToolConfigOptions &\n GeminiSafetyOptions &\n GeminiCommonConfigOptions &\n GeminiCachedContentOptions &\n GeminiStructuredOutputOptions\n}\n\n/**\n * Type-only map from chat model name to its supported input modalities.\n * Based on the 'supports.input' arrays defined for each model.\n * Note: 'document' in the model meta is mapped to 'document' modality.\n * Used by the core AI types to constrain ContentPart types based on the selected model.\n * Note: These must be inlined as readonly arrays (not typeof) because the model\n * constants are not exported and typeof references don't work in .d.ts files\n * when consumed by external packages.\n *\n * @see https://ai.google.dev/gemini-api/docs/vision\n * @see https://ai.google.dev/gemini-api/docs/audio\n * @see https://ai.google.dev/gemini-api/docs/document-processing\n */\nexport type GeminiModelInputModalitiesByName = {\n // Models with full multimodal support (text, image, audio, video, document)\n [GEMINI_3_PRO.name]: typeof GEMINI_3_PRO.supports.input\n [GEMINI_3_FLASH.name]: typeof GEMINI_3_FLASH.supports.input\n [GEMINI_2_5_PRO.name]: typeof GEMINI_2_5_PRO.supports.input\n [GEMINI_2_5_FLASH_LITE.name]: typeof GEMINI_2_5_FLASH_LITE.supports.input\n [GEMINI_2_5_FLASH_LITE_PREVIEW.name]: typeof GEMINI_2_5_FLASH_LITE_PREVIEW.supports.input\n\n // Models with text, image, audio, video (no document)\n [GEMINI_2_5_FLASH.name]: typeof GEMINI_2_5_FLASH.supports.input\n [GEMINI_2_5_FLASH_PREVIEW.name]: typeof GEMINI_2_5_FLASH_PREVIEW.supports.input\n [GEMINI_2_FLASH.name]: typeof GEMINI_2_FLASH.supports.input\n [GEMINI_2_FLASH_LITE.name]: typeof GEMINI_2_FLASH_LITE.supports.input\n}\n"],"names":[],"mappings":"AAiDA,MAAM,eAAe;AAAA,EACnB,MAAM;AA2BR;AAUA,MAAM,iBAAiB;AAAA,EACrB,MAAM;AA2BR;AAUA,MAAM,qBAAqB;AAAA,EACzB,MAAM;AAuBR;AAUA,MAAM,iBAAiB;AAAA,EACrB,MAAM;AA4BR;AASA,MAAM,qBAAqB;AAAA,EACzB,MAAM;AAiBR;AAOA,MAAM,mBAAmB;AAAA,EACvB,MAAM;AA4BR;AASA,MAAM,2BAA2B;AAAA,EAC/B,MAAM;AA2BR;AASA,MAAM,yBAAyB;AAAA,EAC7B,MAAM;AAuBR;AAyCA,MAAM,uBAAuB;AAAA,EAC3B,MAAM;AAiBR;AAOA,MAAM,wBAAwB;AAAA,EAC5B,MAAM;AA2BR;AASA,MAAM,gCAAgC;AAAA,EACpC,MAAM;AA0BR;AASA,MAAM,iBAAiB;AAAA,EACrB,MAAM;AA0BR;AAQA,MAAM,uBAAuB;AAAA,EAC3B,MAAM;AAsBR;AAyCA,MAAM,sBAAsB;AAAA,EAC1B,MAAM;AAsBR;AAQA,MAAM,oBAAoB;AAAA,EACxB,MAAM;AAeR;AAOA,MAAM,0BAA0B;AAAA,EAC9B,MAAM;AAeR;AAOA,MAAM,yBAAyB;AAAA,EAC7B,MAAM;AAeR;AAOA,MAAM,WAAW;AAAA,EACf,MAAM;AAcR;AAmJO,MAAM,gBAAgB;AAAA,EAC3B,aAAa;AAAA,EACb,eAAe;AAAA,EACf,eAAe;AAAA,EACf,iBAAiB;AAAA,EACjB,yBAAyB;AAAA,EACzB,sBAAsB;AAAA,EACtB,8BAA8B;AAAA,EAC9B,eAAe;AAAA,EACf,oBAAoB;AACtB;AAMO,MAAM,sBAAsB;AAAA,EACjC,mBAAmB;AAAA,EACnB,uBAAuB;AAAA,EACvB,qBAAqB;AAAA,EACrB,SAAS;AAAA,EACT,kBAAkB;AAAA,EAClB,uBAAuB;AAAA,EACvB,wBAAwB;AAC1B;AAMO,MAAM,oBAAoB;AAAA,EAC/B,qBAAqB;AAAA,EACrB,mBAAmB;AACrB;AAMO,MAAM,oBAAoB;AAAA,EAC/B;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AACF;"}
@@ -13,104 +13,102 @@ export interface GeminiSafetyOptions {
13
13
  */
14
14
  safetySettings?: Array<SafetySetting>;
15
15
  }
16
- export interface GeminiGenerationConfigOptions {
16
+ export interface GeminiCommonConfigOptions {
17
17
  /**
18
18
  * Configuration options for model generation and outputs.
19
19
  */
20
- generationConfig?: {
21
- /**
22
- * The set of character sequences (up to 5) that will stop output generation. If specified, the API will stop at the first appearance of a stop_sequence. The stop sequence will not be included as part of the response.
23
- */
24
- stopSequences?: Array<string>;
25
- /**
26
- * The requested modalities of the response. Represents the set of modalities that the model can return, and should be expected in the response. This is an exact match to the modalities of the response.
27
-
28
- A model may have multiple combinations of supported modalities. If the requested modalities do not match any of the supported combinations, an error will be returned.
29
- */
30
- responseModalities?: Array<'MODALITY_UNSPECIFIED' | 'TEXT' | 'IMAGE' | 'AUDIO'>;
31
- /**
32
- * Number of generated responses to return. If unset, this will default to 1. Please note that this doesn't work for previous generation models (Gemini 1.0 family)
33
- */
34
- candidateCount?: number;
35
- /**
36
- * The maximum number of tokens to consider when sampling.
37
-
38
- Gemini models use Top-p (nucleus) sampling or a combination of Top-k and nucleus sampling. Top-k sampling considers the set of topK most probable tokens. Models running with nucleus sampling don't allow topK setting.
39
-
40
- Note: The default value varies by Model and is specified by theModel.top_p attribute returned from the getModel function. An empty topK attribute indicates that the model doesn't apply top-k sampling and doesn't allow setting topK on requests.
41
- */
42
- topK?: number;
43
- /**
44
- * Seed used in decoding. If not set, the request uses a randomly generated seed.
45
- */
46
- seed?: number;
47
- /**
48
- * Presence penalty applied to the next token's logprobs if the token has already been seen in the response.
49
-
50
- This penalty is binary on/off and not dependant on the number of times the token is used (after the first). Use frequencyPenalty for a penalty that increases with each use.
51
-
52
- A positive penalty will discourage the use of tokens that have already been used in the response, increasing the vocabulary.
53
-
54
- A negative penalty will encourage the use of tokens that have already been used in the response, decreasing the vocabulary.
55
- */
56
- presencePenalty?: number;
57
- /**
58
- * Frequency penalty applied to the next token's logprobs, multiplied by the number of times each token has been seen in the respponse so far.
59
-
60
- A positive penalty will discourage the use of tokens that have already been used, proportional to the number of times the token has been used: The more a token is used, the more difficult it is for the model to use that token again increasing the vocabulary of responses.
61
-
62
- Caution: A negative penalty will encourage the model to reuse tokens proportional to the number of times the token has been used. Small negative values will reduce the vocabulary of a response. Larger negative values will cause the model to start repeating a common token until it hits the maxOutputTokens limit.
63
- */
64
- frequencyPenalty?: number;
65
- /**
66
- * If true, export the logprobs results in response.
67
- */
68
- responseLogprobs?: boolean;
69
- /**
70
- * Only valid if responseLogprobs=True. This sets the number of top logprobs to return at each decoding step in the Candidate.logprobs_result. The number must be in the range of [0, 20].
71
- */
72
- logprobs?: number;
73
- /**
74
- * Enables enhanced civic answers. It may not be available for all models.
75
- */
76
- enableEnhancedCivicAnswers?: boolean;
77
- /**
78
- * The speech generation config.
79
- */
80
- speechConfig?: {
81
- voiceConfig: {
82
- prebuiltVoiceConfig: {
83
- voiceName: string;
84
- };
85
- };
86
- multiSpeakerVoiceConfig?: {
87
- speakerVoiceConfigs?: Array<{
88
- speaker: string;
89
- voiceConfig: {
90
- prebuiltVoiceConfig: {
91
- voiceName: string;
92
- };
93
- };
94
- }>;
20
+ /**
21
+ * The set of character sequences (up to 5) that will stop output generation. If specified, the API will stop at the first appearance of a stop_sequence. The stop sequence will not be included as part of the response.
22
+ */
23
+ stopSequences?: Array<string>;
24
+ /**
25
+ * The requested modalities of the response. Represents the set of modalities that the model can return, and should be expected in the response. This is an exact match to the modalities of the response.
26
+
27
+ A model may have multiple combinations of supported modalities. If the requested modalities do not match any of the supported combinations, an error will be returned.
28
+ */
29
+ responseModalities?: Array<'MODALITY_UNSPECIFIED' | 'TEXT' | 'IMAGE' | 'AUDIO'>;
30
+ /**
31
+ * Number of generated responses to return. If unset, this will default to 1. Please note that this doesn't work for previous generation models (Gemini 1.0 family)
32
+ */
33
+ candidateCount?: number;
34
+ /**
35
+ * The maximum number of tokens to consider when sampling.
36
+
37
+ Gemini models use Top-p (nucleus) sampling or a combination of Top-k and nucleus sampling. Top-k sampling considers the set of topK most probable tokens. Models running with nucleus sampling don't allow topK setting.
38
+
39
+ Note: The default value varies by Model and is specified by theModel.top_p attribute returned from the getModel function. An empty topK attribute indicates that the model doesn't apply top-k sampling and doesn't allow setting topK on requests.
40
+ */
41
+ topK?: number;
42
+ /**
43
+ * Seed used in decoding. If not set, the request uses a randomly generated seed.
44
+ */
45
+ seed?: number;
46
+ /**
47
+ * Presence penalty applied to the next token's logprobs if the token has already been seen in the response.
48
+
49
+ This penalty is binary on/off and not dependant on the number of times the token is used (after the first). Use frequencyPenalty for a penalty that increases with each use.
50
+
51
+ A positive penalty will discourage the use of tokens that have already been used in the response, increasing the vocabulary.
52
+
53
+ A negative penalty will encourage the use of tokens that have already been used in the response, decreasing the vocabulary.
54
+ */
55
+ presencePenalty?: number;
56
+ /**
57
+ * Frequency penalty applied to the next token's logprobs, multiplied by the number of times each token has been seen in the respponse so far.
58
+
59
+ A positive penalty will discourage the use of tokens that have already been used, proportional to the number of times the token has been used: The more a token is used, the more difficult it is for the model to use that token again increasing the vocabulary of responses.
60
+
61
+ Caution: A negative penalty will encourage the model to reuse tokens proportional to the number of times the token has been used. Small negative values will reduce the vocabulary of a response. Larger negative values will cause the model to start repeating a common token until it hits the maxOutputTokens limit.
62
+ */
63
+ frequencyPenalty?: number;
64
+ /**
65
+ * If true, export the logprobs results in response.
66
+ */
67
+ responseLogprobs?: boolean;
68
+ /**
69
+ * Only valid if responseLogprobs=True. This sets the number of top logprobs to return at each decoding step in the Candidate.logprobs_result. The number must be in the range of [0, 20].
70
+ */
71
+ logprobs?: number;
72
+ /**
73
+ * Enables enhanced civic answers. It may not be available for all models.
74
+ */
75
+ enableEnhancedCivicAnswers?: boolean;
76
+ /**
77
+ * The speech generation config.
78
+ */
79
+ speechConfig?: {
80
+ voiceConfig: {
81
+ prebuiltVoiceConfig: {
82
+ voiceName: string;
95
83
  };
96
- /**
97
- * Language code (in BCP 47 format, e.g. "en-US") for speech synthesis.
98
-
99
- Valid values are: de-DE, en-AU, en-GB, en-IN, en-US, es-US, fr-FR, hi-IN, pt-BR, ar-XA, es-ES, fr-CA, id-ID, it-IT, ja-JP, tr-TR, vi-VN, bn-IN, gu-IN, kn-IN, ml-IN, mr-IN, ta-IN, te-IN, nl-NL, ko-KR, cmn-CN, pl-PL, ru-RU, and th-TH.
100
- */
101
- languageCode?: 'de-DE' | 'en-AU' | 'en-GB' | 'en-IN' | 'en-US' | 'es-US' | 'fr-FR' | 'hi-IN' | 'pt-BR' | 'ar-XA' | 'es-ES' | 'fr-CA' | 'id-ID' | 'it-IT' | 'ja-JP' | 'tr-TR' | 'vi-VN' | 'bn-IN' | 'gu-IN' | 'kn-IN' | 'ml-IN' | 'mr-IN' | 'ta-IN' | 'te-IN' | 'nl-NL' | 'ko-KR' | 'cmn-CN' | 'pl-PL' | 'ru-RU' | 'th-TH';
102
84
  };
103
- /**
104
- * Config for image generation. An error will be returned if this field is set for models that don't support these config options.
105
- */
106
- imageConfig?: {
107
- aspectRatio?: '1:1' | '2:3' | '3:2' | '3:4' | '4:3' | '9:16' | '16:9' | '21:9';
85
+ multiSpeakerVoiceConfig?: {
86
+ speakerVoiceConfigs?: Array<{
87
+ speaker: string;
88
+ voiceConfig: {
89
+ prebuiltVoiceConfig: {
90
+ voiceName: string;
91
+ };
92
+ };
93
+ }>;
108
94
  };
109
95
  /**
110
- * If specified, the media resolution specified will be used.
111
- */
112
- mediaResolution?: MediaResolution;
113
- } & GeminiThinkingOptions & GeminiStructuredOutputOptions;
96
+ * Language code (in BCP 47 format, e.g. "en-US") for speech synthesis.
97
+
98
+ Valid values are: de-DE, en-AU, en-GB, en-IN, en-US, es-US, fr-FR, hi-IN, pt-BR, ar-XA, es-ES, fr-CA, id-ID, it-IT, ja-JP, tr-TR, vi-VN, bn-IN, gu-IN, kn-IN, ml-IN, mr-IN, ta-IN, te-IN, nl-NL, ko-KR, cmn-CN, pl-PL, ru-RU, and th-TH.
99
+ */
100
+ languageCode?: 'de-DE' | 'en-AU' | 'en-GB' | 'en-IN' | 'en-US' | 'es-US' | 'fr-FR' | 'hi-IN' | 'pt-BR' | 'ar-XA' | 'es-ES' | 'fr-CA' | 'id-ID' | 'it-IT' | 'ja-JP' | 'tr-TR' | 'vi-VN' | 'bn-IN' | 'gu-IN' | 'kn-IN' | 'ml-IN' | 'mr-IN' | 'ta-IN' | 'te-IN' | 'nl-NL' | 'ko-KR' | 'cmn-CN' | 'pl-PL' | 'ru-RU' | 'th-TH';
101
+ };
102
+ /**
103
+ * Config for image generation. An error will be returned if this field is set for models that don't support these config options.
104
+ */
105
+ imageConfig?: {
106
+ aspectRatio?: '1:1' | '2:3' | '3:2' | '3:4' | '4:3' | '9:16' | '16:9' | '21:9';
107
+ };
108
+ /**
109
+ * If specified, the media resolution specified will be used.
110
+ */
111
+ mediaResolution?: MediaResolution;
114
112
  }
115
113
  export interface GeminiCachedContentOptions {
116
114
  /**
@@ -174,11 +172,18 @@ export interface GeminiThinkingOptions {
174
172
  /**
175
173
  * The number of thoughts tokens that the model should generate.
176
174
  */
177
- thinkingBudget: number;
175
+ thinkingBudget?: number;
176
+ };
177
+ }
178
+ export interface GeminiThinkingAdvancedOptions {
179
+ /**
180
+ * Config for thinking features. An error will be returned if this field is set for models that don't support thinking.
181
+ */
182
+ thinkingConfig?: {
178
183
  /**
179
184
  * The level of thoughts tokens that the model should generate.
180
185
  */
181
- thinkingLevel?: ThinkingLevel;
186
+ thinkingLevel?: keyof typeof ThinkingLevel;
182
187
  };
183
188
  }
184
- export type ExternalTextProviderOptions = GeminiToolConfigOptions & GeminiSafetyOptions & GeminiGenerationConfigOptions & GeminiCachedContentOptions;
189
+ export type ExternalTextProviderOptions = GeminiToolConfigOptions & GeminiSafetyOptions & GeminiCommonConfigOptions & GeminiCachedContentOptions & GeminiThinkingOptions & GeminiThinkingAdvancedOptions & GeminiStructuredOutputOptions;
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@tanstack/ai-gemini",
3
- "version": "0.3.1",
3
+ "version": "0.4.0",
4
4
  "description": "Google Gemini adapter for TanStack AI",
5
5
  "author": "",
6
6
  "license": "MIT",
@@ -33,12 +33,12 @@
33
33
  "@google/genai": "^1.30.0"
34
34
  },
35
35
  "peerDependencies": {
36
- "@tanstack/ai": "^0.2.1"
36
+ "@tanstack/ai": "^0.4.0"
37
37
  },
38
38
  "devDependencies": {
39
39
  "@vitest/coverage-v8": "4.0.14",
40
40
  "vite": "^7.2.7",
41
- "@tanstack/ai": "0.2.1"
41
+ "@tanstack/ai": "0.4.0"
42
42
  },
43
43
  "scripts": {
44
44
  "build": "vite build",
@@ -164,13 +164,12 @@ export class GeminiSummarizeAdapter<
164
164
  if (part.text) {
165
165
  accumulatedContent += part.text
166
166
  yield {
167
- type: 'content',
168
- id,
167
+ type: 'TEXT_MESSAGE_CONTENT',
168
+ messageId: id,
169
169
  model,
170
170
  timestamp: Date.now(),
171
171
  delta: part.text,
172
172
  content: accumulatedContent,
173
- role: 'assistant',
174
173
  }
175
174
  }
176
175
  }
@@ -184,8 +183,8 @@ export class GeminiSummarizeAdapter<
184
183
  finishReason === FinishReason.SAFETY
185
184
  ) {
186
185
  yield {
187
- type: 'done',
188
- id,
186
+ type: 'RUN_FINISHED',
187
+ runId: id,
189
188
  model,
190
189
  timestamp: Date.now(),
191
190
  finishReason: