GPT Image 2.5 Flare API

GPT Image 2.5 Flare, OpenAI's newest image model in its fast, balanced tier, via WaveSpeed — 1K images and edits from a reference image. 3¢ an image at medium quality (4¢ for an edit); providerOptions.quality low 1¢, high 9¢, xhigh 16¢, max 36¢ on Varosity Credits.

$0.03/img on Varosity credits.

Use GPT Image 2.5 Flare with your agent Get an API key

no WaveSpeed account needed · provider's list price; credits round each call up to the cent · failed renders never billed

Best for

OpenAI's newest image model
five quality tiers from 1¢
edits from a reference image

Trade-offs

1K output only through Varosity

How to get the best result

GPT Image 2.5 Flare, OpenAI's newest image model in its fast, balanced tier: instruction following and in-image text for everyday images, with a quality dial from 1¢ drafts to 36¢ finals. Pick instead: gpt-image-2.5-sunburst for intricate detail (same price, slower); gpt-image-2 for OpenAI's own endpoint; flux-1-schnell for the cheapest drafts.

Prompting

Order: scene/background, subject, key details, constraints; name the intended use (poster, infographic) to set polish.
Put exact text in quotes with font, color and placement.
For an edit, pass referenceImageUrl and say 'change only X, keep everything else the same'.

Settings on Varosity

Always 1K. REST providerOptions.quality: low 1¢, medium 3¢ (default), high 9¢, xhigh 16¢, max 36¢; an edit with referenceImageUrl adds 1.5¢ before rounding (low 3¢, medium 4¢, high 11¢, xhigh 18¢, max 38¢). Any other quality value is sent and billed as medium.
aspectRatio 1:1, 16:9, 9:16, 4:5 or 21:9; an edit follows the reference image's shape.

Avoid

quality xhigh or max for drafts; settle the layout at low or medium first.
Expecting 2K or 4K output through Varosity.

Example prompt

Poster for a night market. Top center, bold white sans-serif headline "LATE MARKET". Lantern-lit food stalls below, warm light, wet pavement. Footer: "Fridays 8pm".

From wavespeed.ai, wavespeed.ai, checked against what Varosity sends.

Call it from your code

One Varosity key works for every model.

curl -X POST https://varosity.ai/api/v1/images \
  -H "Authorization: Bearer $VAROSITY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "studio product shot of a matte black water bottle, soft rim light",
    "modelId": "gpt-image-2.5-flare",
    "aspectRatio": "1:1"
  }'

Or ask your agent

Connect Claude, Cursor, or any MCP client to https://varosity.ai/api/mcp with your key, then say:

"Use Varosity with gpt-image-2.5-flare to generate an image of …"

Agent setup guide