Skip to content

Models and media ​

Gemini exposes separate model factories for completion, embeddings, two image APIs, and transcription.

Native Gemini images ​

ts
import { GEMINI_2_5_FLASH_IMAGE, GeminiClient } from '@anvia/gemini'

const model = new GeminiClient({ apiKey })
  .imageGenerationModel(GEMINI_2_5_FLASH_IMAGE)

const result = await model.imageGeneration({
  prompt: 'A minimal technical illustration of an agent pipeline',
  width: 1024,
  height: 1024,
})

This path calls generateContent with text and image response modalities. It scans candidate inline parts, decodes all returned images, and rejects invalid or absent base64 data.

Imagen ​

ts
import { IMAGEN_4_GENERATE } from '@anvia/gemini'

const model = gemini.imagenGenerationModel(IMAGEN_4_GENERATE)

Imagen calls generateImages and reads generatedImages[].image.imageBytes. Keep it separate from the native Gemini factory; their request options are not interchangeable.

Both image adapters reduce requested dimensions into an aspect-ratio string. Use additionalParams.config for supported provider controls.

Transcription ​

ts
const transcript = await gemini.transcriptionModel('gemini-2.5-flash')
  .transcription({
    data: audioBytes,
    filename: 'interview.ogg',
    prompt: 'Preserve speaker names and technical terms.',
  })

The adapter adds a transcription-specific system instruction and includes the optional prompt. Supported extension mapping includes WAV, AAC, OGG, FLAC, M4A, and Opus; other filenames default to MPEG.

Model IDs ​

Known unions are autocomplete lists, not an availability guarantee. The broader ModelId type accepts custom strings for new or platform-specific IDs. Use listModels() for inventory and an application allowlist for production selection.

Boundaries ​

The package does not implement text-to-speech as an Anvia audio generation model, even though known completion IDs may include TTS previews. It also does not expose a live audio session. Call a native API directly if the normalized model contract does not fit the desired interaction.

Built for Anvia.