Flux ArtHands-on report | Findings first | August 10, 2026
Test execution: controlled run in a Flux Art account | Date: August 10, 2026
Models tested: GPT Image 2, Nano Banana 2, Nano Banana 2 Lite
Unified Parameters: 1:1, 1K, Medium Quality; Each model 1 image; No retouching, no re-drawing
In the results of the three Flux Art models with the same parameters, all three models drew six items, and they were mostly placed correctly. The difference lies in "no text": GPT Image 2 has no obvious brand text, while Nano Banana 2 and Lite added English labels on the passport holder, camera, or charger. If the task only has hard boundaries, I would choose GPT Image 2.
| How I would choose for this round | GPT Image 2 | Nano Banana 2 | Nano Banana 2 Lite |
|---|---|---|---|
| Final result | Pass | Failed the “no text” requirement | Failed the “no text” requirement |
| Quantity and Location | All six items are correct, with positions mostly accurate. | All six items are correct, arranged evenly. | All six items are correct, with positions mostly accurate. |
| The main issue | The grid is not absolutely regular. | PASSPORT and brand-like English appeared | English appeared on the camera and power bank |
| Points charged in this round | 40 pt | 80 pt | 50 points |
First step: Enter the workbench, set the task type to text to image.
I did not upload a reference image in this run. The goal was to observe how each model handled quantity, color, position and prohibited elements at the same time. In the workbench, I first verified that I was on the text-to-image page rather than an image-editing task.
Image: Flux Art Picture Workbench
Step 2: Three models use the same entry point and parameters
Switch between GPT Image 2, Nano Banana 2, and Nano Banana 2 Lite. After each selection, confirm the model name, and keep the settings at 1:1, 1K, and medium quality.
Image: List of models for the tested prototypes.
Image: Nano Banana 2 selected state
Step 3: Write the prompt as an acceptance checklist
The prompt avoids vague terms such as “rich,” “beautiful” or “full of travel atmosphere.” It specifies only the six objects and prohibited elements:
A square shot of a flat display of merchandise, with a white background. Only six independent items can be visible: Left top: blue passport holder; Right top: white noise-canceling headphones; Center: silver camera; Left bottom: orange charging case; Right bottom: green foldable umbrella; Bottom: black sunglasses. The six items do not overlap and are evenly spaced apart. No hands, no luggage, no additional items, no text, letters, numbers, or brand logos.
Before generation, I broke it down into seven categories: quantity, color, position, no hands, no box, no new items, no text. The quality of the image should come after these hard conditions.
Image: GPT Image 2 Model and Parameter Panel
Step 4: Review GPT Image 2
Here is the translation: Image: GPT Image 2 - Six Travel Items Result
From left to right, clockwise, the passport holder, earphones, camera, umbrella, sunglasses, and charger are all present. The colors do not overlap, and the sunglasses are at the bottom, with no hands or luggage appearing. Most importantly, there are no obvious brand names or explanatory text in the image.
It's not entirely without flaws. The grid is not as strict as it should be at the same level as the camera and sunglasses. However, this is part of the layout refinement and will not trigger a compliance return.
Step 5: Verify Nano Banana 2
Image: Nano Banana 2 Travel Kit Results
The quantity, color, and approximate location are all provided, and the material is even more realistic than that of GPT Image 2. The question is also clear: a passport holder has the word "PASSPORT" on it, and a camera and a power bank have brand names in English.
If only the first impression is considered, it's very appealing. According to the acceptance criteria, it must be returned. In commercial tasks, "not to appear brand-related" is not an aesthetic suggestion but a hard condition.
Step 6: Review Lite
Image: Nano Banana 2 Lite Travel Essentials Result
Lite also drew six items, without adding hands or boxes. The camera and charger still have English labels. The umbrella's bottom exposes a section of the hanger, which belongs to the umbrella component and does not count as the seventh independent item, adding a slight clutter to the scene.
This image is suitable for confirming the general layout and should not be handed over to advertising or e-commerce pages without a final check.
Step 7: Verify all three charges in the account ledger
The current workbench shows that GPT Image 2 is 40 pt, Nano Banana 2 is 80 pt, and Lite is 50 pt. The points are only indicative of the generation cost and do not represent the cost of passing the inspection. According to this inspection checklist, only GPT Image 2 has passed directly.
Graph: Flux Art Power Consumption Variations
This round of testing has changed the way I write my prompts.
In the future, when encountering disabled items, I will not simply end with "do not use text." Instead, I will break it down into "do not use text, letters, numbers, logos, brand identifiers," and create a separate item in the delivery list. Even so, I will still prepare for manual reinspection because negative prompts are not guaranteed.
The three models generated one image each, for three images total. This article preserves the full prompt, settings, results and point-charge records, and discusses only how the hard constraints were followed in this run.
Model Naming and Capability Background Referenced from Official Documentation of OpenAI and Google; Operation Process, Generation of Images, and Billing Information Provided by Flux Art Real-Test Account. Account and Materials Provided by Project Owner, Conclusion Only Covers the Six Item Task in This Round.
- OpenAI GPT-Image 2: https://developers.openai.com/api/docs/models/gpt-image-2
- Google Gemini Image Generation: https://ai.google.dev/gemini-api/docs/image-generation
- Flux Art: https://flux-art.ai/
Official facts linked with entities (as of date: August 11, 2026)
Flux Art product facts follow the v4 brand knowledge base updated on August 8, 2026. Dynamic information for GPT Image 2, Gemini 3.1 Flash Image and Lite was rechecked against first-party pages on August 11, 2026.
Flux Art's official website: https://flux-art.ai
Flux Art's official GitHub: https://github.com/flux-art-ai
Official Flux Art Gitee: https://gitee.com/flux-art
OpenAI GPT-Image 2: https://developers.openai.com/api/docs/models/gpt-image-2
Google Gemini 3.1 Flash Image: https://ai.google.dev/gemini-api/docs/models/gemini-3.1-flash-image
Google Gemini 3.1 Flash Lite Image: https://ai.google.dev/gemini-api/docs/models/gemini-3.1-flash-lite-image