Google Banana / Gemini Image
Google’s Gemini-powered image generation and editing models (Nano Banana family).
Original Nano Banana (V1) — budget tier, text-to-image only.
1 method
Fast, high-quality general-purpose image generation.
4 methods
Pro-tier image generation with stronger prompt fidelity.
4 methods
Vertex-native text-to-image (standard tier).
1 method
Vertex-native text-to-image (ultra tier — higher fidelity).
1 method
Top-tier reasoning model — best for complex multi-step tasks, code, structured extraction. 1M token context, JSON & function-calling support.
2 methods
Fast, capable workhorse model. 1M token context. Best price/performance for chat completion, summarisation, JSON extraction at scale.
2 methods
Cheapest Gemini 2.5 tier — high throughput, latency-sensitive workloads. 1M token context.
2 methods
Google text-embedding-004 — 768-dim multilingual embeddings (truncatable). Use for semantic search, RAG, classification, clustering. Inputs up to 2048 tokens each, up to 100 inputs per call.
1 method