Models

Three models are available, each with text-to-image and image-to-image endpoints.

gemini-2.5-flash-image

Fast model. Output is fixed at 1K.

gemini-3.1-flash-image

Higher quality. Selectable resolution and the full aspect-ratio set.

gemini-3-pro-image

Top tier. Selectable resolution and the full aspect-ratio set.

Endpoints

All image models use the unified jobs endpoint with the model id in the body:

POST /api/v1/generations

Model	Resolution	Notes
gemini-2.5-flash-image	1K only	resolution is ignored (always 1K)
gemini-3.1-flash-image	1K · 2K · 4K	Full aspect-ratio set
gemini-3-pro-image	1K · 2K · 4K	Full aspect-ratio set

The legacy prefixed endpoints (POST /nb/text-generations, POST /nb2/url-generations, etc.) remain available as deprecated aliases — they map to the fixed model of each prefix.

Differences

gemini-2.5-flash-image — fast

Output is fixed at 1K. The resolution field is ignored (the model always outputs 1K). Supported aspect ratios are a subset of the full set — see Aspect ratios.

gemini-3.1-flash-image — higher quality

Selectable resolution (1K, 2K, 4K) and the full aspect-ratio set, including 1:4, 1:8, 4:1, 8:1, and auto.

gemini-3-pro-image — top tier

Selectable resolution (1K, 2K, 4K) and the full aspect-ratio set. Uses the unprefixed endpoints POST /text-generations and POST /url-generations.

Text models

Moonez.ai also serves Gemini text models over a Google-compatible API, billed per token — see Text models for the model list, endpoints, and pricing. Text models are also available through the unified jobs API (POST /api/v1/generations with the model id in the body) — billing is per token in both APIs.

Video models (Veo & Omni)

Veo (veo-3.1-generate-001) generates short videos (MP4, up to 8 seconds) from a text prompt. It is an async job like image generation: submit a prompt, poll GET /jobs/{job_id} until done, then download the result URL (a direct MP4 link) from result.image_url.

See Video generation for the endpoint and request format. Veo is available both through the Gemini-compatible API and the unified jobs API (add duration_seconds to the body).

Gemini Omni (gemini-omni-1.1-flash) is a multimodal real-time model (audio/video input). It is managed in the video category (same routing, limits and pause controls) but is served like a text model, billed per token — through the Gemini-compatible API (generateContent) and through the unified jobs API (POST /api/v1/generations).

Using the Gemini-compatible API

Reference each image model by its Gemini id: gemini-2.5-flash-image, gemini-3.1-flash-image, gemini-3-pro-image. The -preview-suffixed IDs are supported too. Text models use their Gemini IDs directly. See Gemini-compatible API.