AI image generation models: nine that draw instead of describe

Prices from official vendor price lists · October 2026
9 models that create and edit images from a prompt, billed per image or by resolution. Output prices run from $1.5 to $15 per 1M tokens across the 7 models with a paid per-token price; 7 have a free tier.
Image generation
Services found: 9
#AI modelCapabilitiesAPI $/1M (in / out)Context windowPrice / moOutput $/1MOverviewVisit site
1

Nano Banana 2

Google
VisionImage generationStructured JSON output
API onlyView
2

Nano Banana 2 Lite

Google
VisionImage generationStructured JSON output
API onlyView
3

Nano Banana Pro

Google
VisionImage generationStructured JSON output
$2 / $12API only12View
4

Nano Banana 2 (Gemini 3.1 Flash Image)

Google
VisionAdjustable thinkingImage generation
$0.5 / $3131.1K tokensAPI only3View
5

Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image)

Google
VisionAdjustable thinkingImage generation
$0.25 / $1.565.5K tokensAPI only1.5View
6

GPT-5.4 Image 2

OpenAI
VisionImage generationStructured JSON output
$8 / $15272K tokensAPI only15View
7

Nano Banana (Gemini 2.5 Flash Image)

Google
VisionImage generationWeb search grounding
$0.3 / $2.532.8K tokensAPI only2.5View
8

Nano Banana Pro (Gemini 3 Pro Image)

Google
VisionAdjustable thinkingImage generation
$2 / $12131.1K tokensAPI only12View
9

GPT-5 Image Mini

OpenAI
VisionImage generationStructured JSON output
$2.5 / $2400K tokensAPI only2View

Browse by need

Reasoning models169 modelsCoding assistants17 modelsVision and images156 modelsLong context 1M+82 modelsFree tier available80 modelsOpen weights88 models

Nine models in this rating generate and edit images from a text prompt. Google Nano Banana line dominates - Nano Banana 2, 2 Lite and Pro - with output billed from $1.50 to $15 per million tokens, or per image depending on vendor.

Image pricing does not behave like text pricing. Output is charged per image or in image tokens tied to resolution, so a 4K render costs several times a 1K one from the same model. Editing also bills input tokens for the source. The practical questions are resolution, whether the model does true editing rather than regeneration, and whether it renders legible text inside the image.

Resolution is priceOutput bills by image tokens tied to pixels — 4K costs several times 1K.
Editing is not regenerationSome models edit the original; others rebuild it and lose details.
Text remains the weak pointLegible words inside an image are still the most common failure.

Models with this capability

The 9 models below all support this capability. Prices are input and output per 1M tokens from the vendor official price list; for dirhams multiply by 3.6725. Tap a card to open the model page.

Nano Banana 2
#1
Google DeepMind
price on request
Nano Banana 2 Lite
#2
Google DeepMind
price on request
Nano Banana Pro
#3
Google DeepMind
in $2out $12
Nano Banana 2 (Gemini 3.1 Flash Image)
#4
Google DeepMind
in $0.5out $3131.1K tokens
Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image)
#5
Google DeepMind
in $0.25out $1.565.5K tokens
GPT-5.4 Image 2
#6
OpenAI
in $8out $15272K tokens
Nano Banana (Gemini 2.5 Flash Image)
#7
Google DeepMind
in $0.3out $2.532.8K tokens
Nano Banana Pro (Gemini 3 Pro Image)
#8
Google DeepMind
in $2out $12131.1K tokens

How to choose

  • Decide whether you need editing or generation — the models differ.
  • Test typography early if your images carry words.
  • Check commercial-use terms before shipping generated assets.
  • Budget per image, not per token, when comparing vendors.

Common mistakes

Rendering at maximum resolution by defaultMost placements never display the extra pixels.
Assuming consistency across callsCharacter and style drift between generations without careful prompting.

Who this suits

MarketingCampaign visuals and variants without a studio.
ProductPlaceholder and illustrative imagery at scale.
CommerceBackground replacement and catalogue cleanup.

Related filters

Vision156 models that read images
Video understanding53 models that watch and explain video
Frontier tier41 flagship models, the best each vendor sells
MultilingualAll 239 models work beyond English
All 239 modelsThe full rating with filters and prices

Frequently asked questions

Why does a 4K image cost several times a 1K one?

Output is charged in image tokens tied to resolution, so pixels drive price directly.

Can these models edit an existing image?

Some do true editing; others regenerate from scratch. Editing also bills input tokens for the source image.

Do they render text inside images correctly?

Still the most common failure mode. Test with your own copy before relying on it.

Can these models edit photographs?

Yes, though results vary between true editing and regeneration.

Do they keep a consistent character?

Only with deliberate prompting and reference images, and never perfectly.

Frequently searched

By capabilityReasoningCodingVisionLong context 1M+Free tierOpen weightsLow latencyRealtime voiceImage generation
By vendorAnthropic (13)OpenAI (30)Google DeepMind (17)xAI (8)Meta (6)DeepSeek (9)Mistral AI (11)Alibaba (Qwen) (34)Z.ai (12)Moonshot AI (5)
Popular modelsClaude Fable 5.1GPT-6 AstraClaude Opus 5Claude Fable 5GPT-5.6 SolGLM 5.3Grok 4.6Kimi K3GPT-5.6 Terra
By deliveryBatch APIPrompt cachingPriority tierFrontier tierProprietary APITool callingStructured JSONWeb searchMultilingual

Ratings 0

Avg. score 0

Rate this page

Help us improve