Examples for using gpt-image-2 through RunAPI from agent tools or API calls. Copy a prompt, then use it in Claude Code, Codex, Cursor, Windsurf, or your backend.
Professional character reference sheet for an original fantasy RPG character — a young female mage with silver hair and violet eyes wearing an ornate dark cloak with glowing rune patterns. Clean white background. Three-view turnaround showing front, side, and back. Row of facial expression variations: neutral, smiling, angry, surprised. Detailed costume and equipment breakdowns with callout lines. Color palette swatch row at the bottom. Brief world-building notes in clean typography. Organized grid layout, concept art style, high resolution. 16:9.
curl -X POST https://runapi.ai/api/v1/gpt_image_2/text_to_image \
-H "Authorization: Bearer $RUNAPI_KEY" \
-H "Content-Type: application/json" \
--data-binary @- <<'JSON'
{
"model": "gpt-image-2",
"prompt": "Professional character reference sheet for an original fantasy RPG character — a young female mage with silver hair and violet eyes wearing an ornate dark cloak with glowing rune patterns. Clean white background. Three-view turnaround showing front, side, and back. Row of facial expression variations: neutral, smiling, angry, surprised. Detailed costume and equipment breakdowns with callout lines. Color palette swatch row at the bottom. Brief world-building notes in clean typography. Organized grid layout, concept art style, high resolution. 16:9."
}
JSON
Inside a museum exhibit titled 'Ancient Technology: The Desktop Era', a programmer in a glass display case live-demonstrates coding on a CRT monitor while amazed schoolchildren press their faces against the glass. Exhibit placard reads: 'Homo Developerus (c. 2005) - Primitive human using keyboard-based input devices.' A second display case nearby shows a physical book labeled 'Stack Overflow - Print Edition, Vol. 1 of 4,827.' 2D cartoon illustration style, warm museum lighting, humorous and nostalgic tone. 16:9.
curl -X POST https://runapi.ai/api/v1/gpt_image_2/text_to_image \
-H "Authorization: Bearer $RUNAPI_KEY" \
-H "Content-Type: application/json" \
--data-binary @- <<'JSON'
{
"model": "gpt-image-2",
"prompt": "Inside a museum exhibit titled 'Ancient Technology: The Desktop Era', a programmer in a glass display case live-demonstrates coding on a CRT monitor while amazed schoolchildren press their faces against the glass. Exhibit placard reads: 'Homo Developerus (c. 2005) - Primitive human using keyboard-based input devices.' A second display case nearby shows a physical book labeled 'Stack Overflow - Print Edition, Vol. 1 of 4,827.' 2D cartoon illustration style, warm museum lighting, humorous and nostalgic tone. 16:9."
}
JSON
Hand-drawn city food map of Chengdu with a bird's-eye-view illustrated base map, 12 hand-drawn food spot illustrations showing local dishes — mapo tofu, dan dan noodles, hot pot, rabbit heads — with handwritten labels in both Chinese and English, vine and chili pepper decorative border, title 'Chengdu Foodie Walking Map' in rounded hand-lettered typography, watercolor and colored pencil mixed media texture
curl -X POST https://runapi.ai/api/v1/gpt_image_2/text_to_image \
-H "Authorization: Bearer $RUNAPI_KEY" \
-H "Content-Type: application/json" \
--data-binary @- <<'JSON'
{
"model": "gpt-image-2",
"prompt": "Hand-drawn city food map of Chengdu with a bird's-eye-view illustrated base map, 12 hand-drawn food spot illustrations showing local dishes — mapo tofu, dan dan noodles, hot pot, rabbit heads — with handwritten labels in both Chinese and English, vine and chili pepper decorative border, title 'Chengdu Foodie Walking Map' in rounded hand-lettered typography, watercolor and colored pencil mixed media texture"
}
JSON
A four-panel comic strip titled 'MORNING ROUTINE' — Panel 1: alarm clock ringing, character groaning under covers. Panel 2: character makes coffee with bleary eyes. Panel 3: character sits down at desk looking determined. Panel 4: character already fast asleep at the desk, coffee untouched. Warm minimal illustration style with expressive character design, readable title text, consistent character across all panels. 16:9.
curl -X POST https://runapi.ai/api/v1/gpt_image_2/text_to_image \
-H "Authorization: Bearer $RUNAPI_KEY" \
-H "Content-Type: application/json" \
--data-binary @- <<'JSON'
{
"model": "gpt-image-2",
"prompt": "A four-panel comic strip titled 'MORNING ROUTINE' — Panel 1: alarm clock ringing, character groaning under covers. Panel 2: character makes coffee with bleary eyes. Panel 3: character sits down at desk looking determined. Panel 4: character already fast asleep at the desk, coffee untouched. Warm minimal illustration style with expressive character design, readable title text, consistent character across all panels. 16:9."
}
JSON
A two-panel comic about a small robot learning to paint — Panel 1: robot carefully studying a blank canvas with analytical scanning eyes, paint palette in mechanical hand. Panel 2: robot proudly displaying a messy but charming abstract painting with paint splattered on its chassis. Warm studio lighting, expressive robot body language, simple title 'FIRST ATTEMPT' below. 16:9.
curl -X POST https://runapi.ai/api/v1/gpt_image_2/text_to_image \
-H "Authorization: Bearer $RUNAPI_KEY" \
-H "Content-Type: application/json" \
--data-binary @- <<'JSON'
{
"model": "gpt-image-2",
"prompt": "A two-panel comic about a small robot learning to paint — Panel 1: robot carefully studying a blank canvas with analytical scanning eyes, paint palette in mechanical hand. Panel 2: robot proudly displaying a messy but charming abstract painting with paint splattered on its chassis. Warm studio lighting, expressive robot body language, simple title 'FIRST ATTEMPT' below. 16:9."
}
JSON
Wuxia style swordswoman portrait, cold and sharp phoenix eyes, porcelain-white skin, long straight black hair wet and windswept across her face, soaked dark martial arts outfit clinging to her form, ancient sword held at her side with blue qi energy glow emanating from the blade, moonlit bamboo forest background with fog, 8K masterpiece quality, dramatic rim lighting from the moon. 9:16 vertical.
curl -X POST https://runapi.ai/api/v1/gpt_image_2/text_to_image \
-H "Authorization: Bearer $RUNAPI_KEY" \
-H "Content-Type: application/json" \
--data-binary @- <<'JSON'
{
"model": "gpt-image-2",
"prompt": "Wuxia style swordswoman portrait, cold and sharp phoenix eyes, porcelain-white skin, long straight black hair wet and windswept across her face, soaked dark martial arts outfit clinging to her form, ancient sword held at her side with blue qi energy glow emanating from the blade, moonlit bamboo forest background with fog, 8K masterpiece quality, dramatic rim lighting from the moon. 9:16 vertical."
}
JSON
FAQ
Using gpt-image-2 prompts
What is %{model}?
%{model} is available through RunAPI as part of the unified model catalog. These prompts show practical input patterns that agents and backend services can reuse.
How do I use these prompts?
Copy any prompt and paste it into Claude Code, Codex, Cursor, or Windsurf after installing the RunAPI MCP Server. Developers can also copy the API example and send the prompt directly.
Do these prompts cost money to browse?
Browsing and copying prompt examples is free. Generation requests only cost money when you call a RunAPI model with your API key.
Can I adapt the prompts for production?
Yes. Treat each prompt as a starting point, then add your brand rules, output dimensions, safety constraints, and application-specific context before using it in production.