innFactory AI Consulting from Rosenheim advises companies in the DACH region on GDPR-compliant integration of image generation AI. Gemini 2.5 Flash Image (gemini-2.5-flash-image, “Nano Banana”) is Google’s first Gemini image model in general availability (since 2 October 2025) and is deprecated: per Google it is shut down on 15 March 2027. Google’s recommended successor is Gemini 3.1 Flash Lite Image (Nano Banana 2 Lite, gemini-3.1-flash-lite-image); newer and more capable models are Nano Banana 2.1 and Nano Banana Pro, see Gemini 3 Pro Image and the Nano Banana family. As of October 2026.
Current models at a glance
| Model | API ID | Status | Region on the Agent Platform | Price (Gemini API) |
|---|---|---|---|---|
| Gemini 2.5 Flash Image (Nano Banana) | gemini-2.5-flash-image | deprecated, retirement 15 March 2027 | global, EU single regions, US regions | USD 0.30 input per 1M tokens, USD 0.039 per image |
| Gemini 3.1 Flash Lite Image (Nano Banana 2 Lite) | gemini-3.1-flash-lite-image | GA, recommended replacement | global only | USD 0.0336 per 1K image |
| Gemini Nano Banana 2.1 | gemini-nano-banana-2.1 | GA since 6 October 2026 | global only | USD 0.0336 (1K), USD 0.0504 (2K), USD 0.113 (4K) |
Gemini 2.5 Flash Image
Gemini 2.5 Flash Image is a fast, cost-efficient image model of the Gemini 2.5 family. It generates images from text and edits existing images conversationally: iterative adjustments in natural language, inpainting, adding or removing objects, colourising black-and-white images and combining several images into one composition. Per its model page it takes up to 3 input images per prompt and supports ten aspect ratios from 1:1 to 21:9. The model serves as a reference for existing integrations; we no longer recommend it for new projects.
Pricing
Per the Gemini API pricing page:
| Model | Input (per 1M tokens) | Output |
|---|---|---|
| Gemini 2.5 Flash Image | USD 0.30 (text and image) | USD 0.039 per image |
| Gemini 2.5 Flash Image (batch) | USD 0.15 | USD 0.0195 per image |
| Gemini 3.1 Flash Lite Image | USD 0.25 | USD 0.0336 per 1K image |
| Gemini Nano Banana 2.1 | USD 1.50 | USD 30 per 1M image output tokens (USD 0.0336 per 1K image) |
EU availability and GDPR
| Model | Access route | Regions per model page |
|---|---|---|
| Gemini 2.5 Flash Image | Gemini Enterprise Agent Platform | global; EU single regions europe-west1, europe-west4, europe-north1, europe-central2, europe-west8, europe-southwest1; US regions; ML processing additionally Europe multi-region |
| Gemini 3.1 Flash Lite Image | Gemini Enterprise Agent Platform | global only |
| Gemini Nano Banana 2.1 | Gemini Enterprise Agent Platform | global only |
Frankfurt (europe-west3), London, Zurich and Paris are not listed for Gemini 2.5 Flash Image. Anyone who needs EU residency finds no direct replacement among the successors: Nano Banana 2 Lite and Nano Banana 2.1 run on the Agent Platform in global only. Among the current Nano Banana models, only Nano Banana 2 (gemini-3.1-flash-image) lists EU and US multi-region endpoints, see the Nano Banana overview. For GDPR-critical deployments we also recommend VPC Service Controls and CMEK.
Migration and breaking changes
- Replacement per Google:
gemini-3.1-flash-lite-image(Gemini API and Agent Platform). - Input images: Nano Banana 2 Lite accepts up to 14 input images per prompt instead of 3.
- Resolution: Per its model page, Nano Banana 2 Lite outputs 1K only; 2K or 4K require Nano Banana 2.1 or Pro.
- Aspect ratios: In addition to the previous ten, 1:4, 4:1, 1:8 and 8:1 are possible.
- Region: The successor runs in
globalonly; those who relied on EU single regions must reassess data residency.
Model lifecycle
| Model | Status | Date |
|---|---|---|
gemini-2.5-flash-image | deprecated, retirement | 15 March 2027 (Gemini API deprecations page and Agent Platform model page) |
gemini-2.5-flash-image-preview | shut down | 15 January 2026 |
gemini-2.0-flash-preview-image-generation | shut down | 14 November 2025 |
A warning on the Gemini API pricing page names 2 October 2026 instead; the deprecations page and the model page, with 15 March 2027, are authoritative. Google states retirement dates as earliest possible dates.
Integration with CompanyGPT
For companies that want to integrate image generation into internal workflows, the combination with CompanyGPT works well: employees access image functions through a protected interface, and prompts and results can be steered centrally. The innFactory AI Gateway adds budgets and access control across several image models. For new integrations we recommend one of the successors instead of Gemini 2.5 Flash Image.
Our recommendation
- New projects: do not build on Gemini 2.5 Flash Image; go straight to Nano Banana 2.1 (quality, 1K to 4K), Nano Banana 2 Lite (low cost, 1K) or Nano Banana Pro (for demanding quality requirements, 1K to 4K), see the Nano Banana overview.
- Existing integrations: move to
gemini-3.1-flash-lite-imageor a more capable successor before 15 March 2027 and test prompts and output formats. - EU single region mandatory (for example Belgium or the Netherlands): the Agent Platform lists only
globalfor the successor; data residency must be reassessed, and Nano Banana 2 with EU multi-region may be an option.
We are happy to review the GDPR-compliant integration of image generation into your workflows. Contact us.