---
title: "Which model for what"
description: "Pick the right model for the job before you write a word."
section: "Models"
url: "https://batchbanana.com/docs/models/which-model"
markdown_url: "https://batchbanana.com/docs/models/which-model.md"
last_verified: "2026-10-03"
---

# Which model for what

Pick the right model for the job before you write a word.

## Overview

Each model is good at different things. Some set text accurately, some are better at light and photographs, and two make vector files. Starting on the right model saves you credits and rewrites.

Start with the table of first picks below. Then compare five model families on the same six prompts, and see art from a game made with several models. Prices and the full list of models are in [Models](https://batchbanana.com/docs/models/models).

## Start here

| Job                                  | First pick                                                 | Also good                                        |
| ------------------------------------ | ---------------------------------------------------------- | ------------------------------------------------ |
| Your product in a new scene          | [Nano Banana 2](https://batchbanana.com/docs/models/nano-banana)                  | [GPT Image 2.5 Sunburst](https://batchbanana.com/docs/models/gpt-image) |
| The same character in new scenes     | [GPT Image 2.5 Sunburst](https://batchbanana.com/docs/models/gpt-image)           | [Nano Banana 2](https://batchbanana.com/docs/models/nano-banana)        |
| Exact text, slides and infographics  | [GPT Image 2.5](https://batchbanana.com/docs/models/gpt-image)                    | [Nano Banana Pro](https://batchbanana.com/docs/models/nano-banana)      |
| Posters, logos and packaging         | [Ideogram 4.5](https://batchbanana.com/docs/models/ideogram)                      | [GPT Image 2.5](https://batchbanana.com/docs/models/gpt-image)          |
| Cinematic photos and portraits       | [Seedream 4.5](https://batchbanana.com/docs/models/seedream)                      | [Nano Banana Pro](https://batchbanana.com/docs/models/nano-banana)      |
| Placing things exactly on the canvas | [Seedream 5 Pro](https://batchbanana.com/docs/models/seedream)                    | [Ideogram 4.5](https://batchbanana.com/docs/models/ideogram)            |
| Icons, marks, anything you'll scale  | [Arrow 2](https://batchbanana.com/docs/models/arrow)                              | [Recraft V4.1 SVG](https://batchbanana.com/docs/models/recraft)         |
| A real place or product, by name     | [Flux 3 Image](https://batchbanana.com/docs/models/flux) with **Web search**      | None; no other model searches the web            |
| Cheap drafts                         | [GPT Image 2.5](https://batchbanana.com/docs/models/gpt-image) on **Low** quality | [Nano Banana 2 Lite](https://batchbanana.com/docs/models/nano-banana)   |

These picks are our judgement. They come from what each maker's guide says its model is for, and from one run of the comparison below, not from a benchmark. Your prompt may disagree with them. When you're unsure, add two or three models to a [variety pack](https://batchbanana.com/docs/create/create-images#get-more-range) and run the prompt once on each.

## What every model agrees on

These four rules come up again and again in the makers' guides, and they work on every model here.

- **Quote the exact words.** Any text that must appear goes in quotes, with where it sits and how it looks.
- **Say what the image is for.** A logo for a small chocolate maker, a slide for a pitch, a sticker for a shop. The purpose guides the model's choices of style, layout and detail.
- **Give each reference a job.** Say what to take from each image you attach: the character, the palette, the background.
- **When you change an image, say what changes and what stays.** List the parts that must not move.

Each guide has the details for its model, with examples.

## Same prompt, five models

We ran six prompts and made two images on each family. Nano Banana runs on Nano Banana 2 and Seedream on 5 Pro, both at 2K; GPT Image runs on Sunburst and Ideogram on 4.5, both on High. We show the first image of the two, or the second where we liked it more. Nothing is retouched.

### Product shot

![Halvard bar on an open atlas of Norway by a window, HALVARD large on the cobalt wrapper and a small added DARK 72% label](https://batchbanana.com/docs/models/hub/product-shot-nano-banana.webp)

*Nano Banana 2*

![Halvard bar on an atlas of Norway beside a coffee cup and compass, the wordmark printed with a stray dot as HALVAR.D, silver foil showing at the open end](https://batchbanana.com/docs/models/hub/product-shot-gpt-image.webp)

*GPT Image 2.5 Sunburst*

![Halvard bar lying diagonally across an atlas of Norway, HALVARD running along the cobalt wrapper, the chocolate end showing](https://batchbanana.com/docs/models/hub/product-shot-ideogram.webp)

*Ideogram 4.5*

![Halvard bar on an atlas in a band of hard sunlight, a large cream HALVARD on the contour-lined wrapper, three squares of chocolate showing](https://batchbanana.com/docs/models/hub/product-shot-seedream.webp)

*Seedream 5 Pro*

![Flat illustration of the Halvard bar on a beige map marked NORGE, the cobalt wrapper with white contour lines](https://batchbanana.com/docs/models/hub/product-shot-arrow.svg)

*Arrow 2*

*Prompt for all 5: "A Halvard chocolate bar lying across an old atlas open at the coast of Norway, hard window light. Its uncoated cobalt wrapper is printed with fine white contour lines and "HALVARD" in a tall cream serif, cut short so the last squares of dark chocolate show."*

GPT Image put a stray dot before the D in "HALVARD"; the other four set it clean. Nano Banana 2 and GPT Image both added small print nobody asked for.

### Game asset

This row is not one prompt on five models. It is art from [Geolingo](https://geolingo.app/), a geography game we built, and every asset in the game was made in Batch Banana.

**Nano Banana 2**

![Game icon of Big Ben in warm stone with a dark slate spire and two clock faces, on plain cream](https://batchbanana.com/docs/models/hub/game-asset-big-ben.webp)

*Prompt: "Elizabeth Tower (Big Ben), London, as a single standalone landmark emblem for the Western Europe stop: a slender Gothic clock tower with its recognizable clock face, upper pinnacles and pointed roof, in gentle three-quarter view." Model: Nano Banana 2.*

**Nano Banana 2 Lite**

![White quiz card on a stack of red-edged cards, a faint map border around the text Starts With C: North America, 3 countries in North America start with C, and a 3 countries pill](https://batchbanana.com/docs/models/hub/game-asset-quiz-card.webp)

*Prompt: "Create a polished vertical GeoLingo geography quiz card template for a mobile app, using Image 1 as the guide for the card proportions, rounded corners, stacked-card presentation, and centered content area; use Images 2 and 3 only as references for the vintage cartographic border language and small nautical motifs. Redesign the material treatment so the card feels crisp, bright, and freshly printed rather than aged or dirty. The card face is clean soft white with only the faintest warm ivory cast, even illumination, and abundant clear negative space. Paper texture is extremely subtle and fine-grained, with virtually no foxing, stains, brown speckles, grime, yellowed edges, smudges, or distressed patches. Keep the charming vintage travel-map character through precise dark brown linework, muted powder-blue coastline accents, restrained sage green, and small matte golden corner emblems. Use sharp, consistent outlines and clean print registration. Keep decorative maps and routes confined to a narrow perimeter so the central reading area remains uncluttered and highly legible. Include the exact centered copy: “Starts With C: North America” in bold near-black sans-serif, “3 countries in North America start with C.” in medium gray sans-serif, and a small outlined cream pill reading “3 countries”. The typography is crisp, evenly spaced, correctly spelled, and sized for mobile readability. The stacked cards behind it are also clean white with restrained red accents. Overall mood: vintage-inspired but pristine, airy, modern, and app-ready; a freshly manufactured premium card deck, not an antique artifact." Model: Nano Banana 2 Lite.*

**Krea 2**

![Cream card back with a compass rose over a pale world rosette in the center and a small compass medallion in each corner, inside a blue and gold border](https://batchbanana.com/docs/models/hub/game-asset-card-back-krea.webp)

*Prompt: "Design a crafted GeoLingo card back centered on a geometric cartographic rosette. Combine a simplified compass rose, curved latitude lines, and four abstract coastline shapes into one unified circular emblem. The coastlines should feel like fragments of a map without depicting specific countries. Repeat a smaller portion of the emblem in each corner. Use strict four-way symmetry, clean spacing, and a strong hierarchy: central rosette first, border second, texture last. Use muted sky blue, warm cream, mustard gold, brick-red accents, and dark-brown linework. Match the attached card front’s restrained vintage print character. No scattered travel icons, literal world map, pins, landmarks, text, shadows, perspective, or excessive weathering." Model: Krea 2.*

**Nano Banana Pro**

![Brick-red card back with a smiling cartoon globe holding a gold card in the top-left corner, the same globe upside down in the bottom-right, and a small cream diamond in the middle](https://batchbanana.com/docs/models/hub/game-asset-card-back-pro.webp)

*Prompt: "Create a minimal traditional card back for the GeoLingo Country Rush game, using Image 1 for its bold red geometry and Image 2 for the recognizable mascot. Use a deep brick-red field with two broad cream quarter-circle bands in opposite corners. Crop one large smiling globe mascot into the upper-left corner and repeat it upside down in the lower-right, so each character partially follows the curve of the bands. Give each mascot one visible white-gloved hand presenting a single oversized gold-backed card toward the center. Leave the center mostly open except for one small cream diamond. Add one thick cream border. Perfect rotational symmetry, large flat shapes, strong dark outlines, minimal facial and map detail, clear at mobile thumbnail size. No additional stars, tiny motifs, text, or mockup; show the complete card back straight on." Model: Nano Banana Pro.*

**GPT Image 2.5 Sunburst**

![Wide banner with North America in blue script on cream, pine forest, mountains and a lake on the left, a stepped pyramid, agave and palms on the right](https://batchbanana.com/docs/models/hub/game-asset-journey-header.webp)

*Prompt: "Create one standalone 21:9 GeoLingo North America Journey header using Image 1’s centered-title and separated-side-scenery composition. Stack the exact title on two centered lines: “North” above “America”. Match the flowing stacked script structure of “South America” in Image 3, but use a muted dusty-blue fill, near-black outline, and compact crisp dark-brown shadow. The title occupies about 30–34% of the width. Left cluster: Canadian Rockies, pine forest, small dusty-blue lake. Right cluster: Mexican stepped pyramid, agave, Caribbean palms and coastline. Avoid USA monuments. Keep the center open and never overlap the title. Every scenic element finishes above a clean 12–15% warm-cream bottom safe area; no bottom crop, base strip, or cutoff. Flat hard-edged GeoLingo illustration, matte cream, dusty blue, sage, tan, faded red. No character, weather, progression, frame, gradient, blur, extra text, or watermark." Model: GPT Image 2.5 Sunburst.*

Nano Banana 2 made about 2,100 of the project's 2,800 files, including every landmark icon. GPT Image 2 made icon sheets, treasure chests and paywall drafts. Krea 2 and Nano Banana Pro made card backs, and Ideogram 4 a region wordmark. Each asset used earlier Geolingo art as a style reference, which this page does not show. The Big Ben prompt is cut to its subject; the full prompt attached nine style images.

### Poster with text

This row did not use one shared prompt. We wrote one brief, and Agent adapted it into a prompt for each model, so each image opens its own prompt. Every prompt mentions a style reference, Image 1, which this page does not show.

**Nano Banana 2**

![Fresco poster with MORE in tall condensed white letters over slate-blue clouds, BATCH BANANA MODEL SHOWCASE on one line just above it, figures crowding the lower half](https://batchbanana.com/docs/models/hub/more-poster-nano-banana.webp)

*Prompt: "A bold editorial poster for the Batch Banana docs model showcase. Renaissance ceiling-fresco figures drift and reach toward one another through stormy slate-blue clouds, painted in warm umber, cream, and faded terracotta with aged-paper grain. Across the middle, a monumental custom high-contrast white display word reads "MORE", its letterforms cutting sharply across the painted scene. Small restrained uppercase sans-serif text at the top reads "BATCH BANANA" and "MODEL SHOWCASE"; a small line at the bottom reads "ONE PROMPT. MANY POSSIBILITIES." Image 1 is the aesthetic reference: take its daring scale contrast, celestial Renaissance atmosphere, muted palette, and refined editorial typography, not its actual words or figures. Make the composition striking and legible at docs thumbnail size." Model: Nano Banana 2.*

**GPT Image 2.5 Sunburst**

![Fresco poster with MORE in a white serif crossed by a hairline with small stars, two pairs of figures reaching toward each other above and below it](https://batchbanana.com/docs/models/hub/more-poster-gpt-image.webp)

*Prompt: "A bold editorial poster for the Batch Banana docs model showcase. Scene: Renaissance ceiling-fresco figures drift and reach toward one another through stormy slate-blue clouds, painted in warm umber, cream, and faded terracotta with aged-paper grain. Typography: across the middle, one monumental custom high-contrast white display word reads "MORE", cutting sharply across the painted scene. Small uppercase sans-serif text at the upper left reads "BATCH BANANA" once; at the upper right, "MODEL SHOWCASE" once. At the bottom, a small line reads "ONE PROMPT. MANY POSSIBILITIES." once. Image 1 is the aesthetic reference: use its daring type scale, celestial Renaissance atmosphere, muted palette, and editorial restraint, but not its actual words or figures. The headline must remain legible at docs thumbnail size. No other text or watermark." Model: GPT Image 2.5 Sunburst.*

**Ideogram 4.5**

![Fresco poster crowded with figures on grey clouds under a gold sky, MORE in white serif letters spanning the full width](https://batchbanana.com/docs/models/hub/more-poster-ideogram.webp)

*Prompt: "Editorial poster for the Batch Banana docs model showcase. A monumental white high-contrast display headline "MORE" spans the center, perfectly readable against a celestial Renaissance fresco of figures reaching through slate-blue clouds. Muted umber, terracotta, cream, aged-paper grain; daring editorial typography and dramatic scale. Small uppercase sans-serif "BATCH BANANA" at top left, "MODEL SHOWCASE" at top right, and "ONE PROMPT. MANY POSSIBILITIES." at bottom. Image 1 supplies the bold type-to-image hierarchy, fresco atmosphere, and muted palette, not its actual words or figures. Polished graphic design, legible at thumbnail size." Model: Ideogram 4.5.*

**Seedream 5 Pro**

![Fresco poster with MORE in a white serif crossed by a hairline with small stars, a winged figure above and bearded figures in the clouds below](https://batchbanana.com/docs/models/hub/more-poster-seedream.webp)

*Prompt: "A bold editorial poster for the Batch Banana docs model showcase: Renaissance fresco figures reach through stormy slate-blue clouds behind the monumental white display word "MORE" across the center <bbox>75 330 925 670</bbox>. Small uppercase sans-serif "BATCH BANANA" at the upper left, "MODEL SHOWCASE" at the upper right, and "ONE PROMPT. MANY POSSIBILITIES." along the bottom. Paint the figures in warm umber, cream, and faded terracotta, with subtle aged-paper grain. Image 1 guides the daring typography scale, celestial fresco mood, and muted palette, not its words or specific figures. Clear hierarchy and strong thumbnail legibility." Model: Seedream 5 Pro.*

All four spelled every line right, and all four put MORE across the middle. GPT Image, Ideogram and Seedream set BATCH BANANA and MODEL SHOWCASE in the top corners. Nano Banana 2 ran them together as one line just above MORE. GPT Image and Seedream drew nearly the same headline: a tall serif crossed by a hairline with small four-pointed stars. GPT Image arranged the fresco as two pairs of figures reaching toward each other across the word. Ideogram filled the whole sheet with figures on grey clouds under a gold sky. Nano Banana 2 and Seedream kept open blue sky around the word and put most figures below it. The infographic with labels, at the end of this page, is a harder test of text.

### Character from a reference

![Watercolor of Fern on a crate at a lantern-lit night market reading a You Are Here map, badgers and an owl shopping behind](https://batchbanana.com/docs/models/hub/character-from-reference-nano-banana.webp)

*Nano Banana 2*

![Painterly scene of Fern perched on a wooden crate under a paper lantern at a night market, reading a map](https://batchbanana.com/docs/models/hub/character-from-reference-gpt-image.webp)

*GPT Image 2.5 Sunburst*

![Fern reading a large paper map on a broken crate beneath a lantern in a market lane at dusk](https://batchbanana.com/docs/models/hub/character-from-reference-ideogram.webp)

*Ideogram 4.5*

![Fern on a crate at a busy night market reading a map, a paper lantern glowing on the cobbles beside the crate](https://batchbanana.com/docs/models/hub/character-from-reference-seedream.webp)

*Seedream 5 Pro*

![Flat vector of Fern on a crate in the olive coat and satchel, reading a map under a paper lantern, the market in dark blue shapes](https://batchbanana.com/docs/models/hub/character-from-reference-arrow.svg)

*Arrow 2*

*Prompt for all 5: "Fern, the hare in Image 1, same face, coat, satchel, colours and ink-and-watercolour style, sitting on an upturned crate at a night market, reading a paper map by the light of a paper lantern."*

Each model got the same reference, Fern the hare in an olive coat, made for these guides. All five kept the coat and the satchel. Nano Banana 2 filled the market with badgers and an owl.

### Icon

![Pixel-art knight holding a glowing lantern that breaks the top edge of a violet square, a ruined wall behind](https://batchbanana.com/docs/models/hub/icon-nano-banana.webp)

*Nano Banana 2*

![Rounded app icon of a pixel knight with a red plume raising a lantern past the top edge, castle towers at night behind](https://batchbanana.com/docs/models/hub/icon-gpt-image.webp)

*GPT Image 2.5 Sunburst*

![Grey pixel knight holding a lantern above its head inside a plain dark magenta square](https://batchbanana.com/docs/models/hub/icon-ideogram.webp)

*Ideogram 4.5*

![Rounded app icon of a pixel knight with a flail raising a lantern above a violet castle, the lantern breaking the top edge](https://batchbanana.com/docs/models/hub/icon-seedream.webp)

*Seedream 5 Pro*

![Chunky pixel knight raising a lantern that breaks the top edge of a rounded violet tile](https://batchbanana.com/docs/models/hub/icon-arrow.svg)

*Arrow 2*

*Prompt for all 5: "An app icon for Lantern Keep, a pixel-art castle game: a small armored knight holding up a glowing lantern on a deep violet square, the lantern breaking the top edge. Chunky pixel art, thick dark outline, no text."*

Ideogram kept the lantern inside the square, so nothing breaks the top edge.

### Photo

![Raven on a frosted fence post at dawn, its breath visible, a pale sunrise over a frozen field](https://batchbanana.com/docs/models/hub/photo-nano-banana.webp)

*Nano Banana 2*

![Close photo of a raven on a frosted post, violet sheen on its feathers and breath in the cold air](https://batchbanana.com/docs/models/hub/photo-gpt-image.webp)

*GPT Image 2.5 Sunburst*

![Raven facing the camera on a frosted fence post, a soft grey-green field behind](https://batchbanana.com/docs/models/hub/photo-ideogram.webp)

*Ideogram 4.5*

![Raven in profile on a fence post above barbed wire, its breath drifting, a misty grey field behind](https://batchbanana.com/docs/models/hub/photo-seedream.webp)

*Seedream 5 Pro*

![Flat vector illustration of a raven on a fence post with frost, in blues and violets](https://batchbanana.com/docs/models/hub/photo-arrow.svg)

*Arrow 2*

*Prompt for all 5: "A raven on a frosted fence post at dawn, its black feathers showing a blue and violet sheen, its breath visible, a pale grey field behind. 400 mm lens."*

Arrow makes vectors only, so it drew a flat illustration instead of a photo.

### Illustration

![Cel anime girl in a red helmet riding a red scooter stacked with pizza boxes along a wet coast road as lightning strikes the sea](https://batchbanana.com/docs/models/hub/illustration-nano-banana.webp)

*Nano Banana 2*

![Cel anime girl on a red pizza scooter splashing through rain on a cliff road, lightning over the sea behind](https://batchbanana.com/docs/models/hub/illustration-gpt-image.webp)

*GPT Image 2.5 Sunburst*

![Cel anime girl on a red scooter with pizza boxes on a dry sunny seafront road, storm clouds gathering ahead](https://batchbanana.com/docs/models/hub/illustration-ideogram.webp)

*Ideogram 4.5*

![Cel anime girl in a red cap on a red pizza scooter in the rain on a seaside street, a lighthouse and lightning behind](https://batchbanana.com/docs/models/hub/illustration-seedream.webp)

*Seedream 5 Pro*

![Flat vector of a rider on a red scooter with pizza boxes on a coast road under a dark blue storm sky, the rider drawn without a head](https://batchbanana.com/docs/models/hub/illustration-arrow.svg)

*Arrow 2*

*Prompt for all 5: "An anime key visual of a girl on a red delivery scooter racing a summer thunderstorm down a seaside street, pizza boxes strapped behind her. 1990s cel anime, hand-painted background, bold ink lines."*

Arrow drew the rider without a head.

### Infographic with labels

![Exploded diagram titled Inside a Halvard bar: a red patterned wrapper, copper foil, dark chocolate, sea salt and rye crumb, each labelled](https://batchbanana.com/docs/models/hub/infographic-with-labels-nano-banana.webp)

*Nano Banana 2*

![Exploded diagram titled Inside a Halvard bar with a printed paper wrapper, crumpled copper foil, dark chocolate, sea salt and rye crumb labelled on the right](https://batchbanana.com/docs/models/hub/infographic-with-labels-gpt-image.webp)

*GPT Image 2.5 Sunburst*

![Exploded diagram titled Inside a Halvard bar with five thin layers, from paper wrapper to rye crumb, each labelled](https://batchbanana.com/docs/models/hub/infographic-with-labels-ideogram.webp)

*Ideogram 4.5*

![Exploded diagram titled Inside a Halvard bar: kraft wrapper, copper foil, dark chocolate, sea salt and rye crumb, labelled on the left](https://batchbanana.com/docs/models/hub/infographic-with-labels-seedream.webp)

*Seedream 5 Pro*

![Flat exploded diagram titled Inside a Halvard bar, two of its five labels misprinted as Corner foil and .7 salt](https://batchbanana.com/docs/models/hub/infographic-with-labels-arrow.svg)

*Arrow 2*

*Prompt for all 5: "A labelled exploded diagram titled "Inside a Halvard bar": a chocolate bar pulled apart into its layers, with five labels on thin lines: "Paper wrapper", "Copper foil", "Dark chocolate", "Sea salt", "Rye crumb". Flat editorial style on cream."*

Every model kept to the five labels. Arrow misprinted two of them, "Corner foil" and ".7 salt".

## Sources

The makers' guides are linked from each model's page.

Images made in Batch Banana on October 2, 2026. Select Prompt on an image to read the prompt and model that made it.

---

Last verified October 3, 2026. Web version: https://batchbanana.com/docs/models/which-model
