Counts the number of tokens in text using Gemini tokenization. Useful for estimating costs, checking input limits, and optimizing prompts before making API calls.

Generates text embeddings using Gemini embedding models. Converts text into numerical vectors for semantic search, similarity comparison, clustering, and classification tasks.

Generates text content or speech audio from prompts using Gemini models. Supports text generation models (Gemini Flash, Pro) and text-to-speech models with configurable parameters. Generated text is nested at results[i].response.data.text. Output may be wrapped in markdown fences (e.g., ```html...```) or preceded by explanatory prose; strip these before file writing or rendering.

Generates one downloadable raster image from a text prompt. Instant uses Nano Banana 2 Lite (gemini-3.1-flash-lite-image), one 1K square image, a 1-1000 character prompt and default settings. BYOK retains the schema's model choices and supported options. A successful call returns data.image; no generated image is an error. Image generation is synchronous, but a transport timeout can leave the provider outcome unknown. Do not automatically retry an ambiguous submission: another generation can incur another charge.

Submits one Veo video generation request and returns operation_name for tracking. Instant supports Veo 3.1 Lite, 720p, an explicit integer duration of 4/6/8 seconds, a 1-1000 character text prompt and aspect ratio 16:9 or 9:16. Pass the returned handle unchanged to GEMINI_GET_VIDEOS_OPERATION for status or GEMINI_WAIT_FOR_VIDEO to retrieve the MP4. Submission success means Google accepted the job, not that a video file is already available. Instant charges the disclosed nonrefundable quote on acceptance; status and result reads are included. A timeout can leave acceptance unknown: do not automatically resubmit. BYOK retains its compatible options and normalization.

Checks an existing Veo operation without waiting for media download. For Instant, response and video_urls are intentionally hidden, including when generation succeeds; read status and next_action. succeeded means retrieve the same operation using GEMINI_WAIT_FOR_VIDEO. pending or unknown means continue polling the same handle, without submitting another paid generation. filtered or failed is a terminal rejection; any resubmission is a separate charged request. For BYOK, response and video_urls retain provider media data. Poll at 10–30 second intervals and allow up to 15 minutes for generation.

Lists available Gemini and Veo models with their capabilities and limits. Useful for discovering supported models and their features before making generation requests.

Polls an existing Veo operation and downloads its MP4 when ready. Instant polls for about 20 seconds per call. pending=true with success=null means this call did not retrieve a terminal result; the job may still be running or may already have completed after a poll timeout. Reuse the exact returned operation_name with this action or GEMINI_GET_VIDEOS_OPERATION; do not resubmit a paid generation to continue. success=true includes video_file, while success=false indicates terminal safety filtering. Provider errors fail the call. BYOK retains its longer polling behaviour. A new generation is a separate request and can incur another charge.