Image GenerationFreemium

ChatGPT Images

AI image generation built into ChatGPT

openai.com
Overview
OpenAI's ChatGPT Images (now powered by GPT Image 2.5, released September 8 2026) is the default image engine across all ChatGPT tiers, replacing both DALL-E 3 and the DALL-E GPT (retired August 30 2026). It generates and precisely edits images from natural language, with up to 4K resolution via the API, a thinking mode for reasoned generation, and new workflow tools including Templates, Sketch, and prompt sharing. Two API variants — Flare (speed-optimized) and Sunburst (detail-optimized) — give developers finer control over cost and fidelity.
Why it matters
ChatGPT Images 2.5 consolidates image generation, editing, and workflow tooling into a surface that already has hundreds of millions of users, displacing standalone tools for a large share of everyday creative tasks. The retirement of the DALL-E GPT and the shift to all-tier access (with undisclosed caps) signals OpenAI treating image gen as a retention feature rather than a upsell lever — worth watching as caps tighten or plans reprice. For API buyers, the Flare/Sunburst split and five quality tiers introduce real cost-optimization handles that earlier GPT Image generations lacked. Lock-in risk is real: conversational edit history and prompt-sharing links are ChatGPT-native and do not export cleanly.
In practice
Marketing teams use Templates and conversational editing to iterate on ad creatives and social assets without leaving ChatGPT, while developers use Flare for high-volume thumbnail generation and Sunburst for precision product-photo edits where subject fidelity matters. The Sketch feature is popular for rapid concept wireframes on mobile — draw a rough layout, describe the output, and get a polished image in seconds. The main honest limitation is that generation caps are opaque: teams doing volume work discover plan limits only after hitting them, and there is no published number to plan against.
Best for
ChatGPT users and developers who need current-generation image generation and precise editing — from quick consumer drafts to API-scale multi-image pipelines — without leaving the chat or managing a separate image tool
Pricing
Image generation available on all ChatGPT plans including Free ($0), Go ($8/mo), Plus ($20/mo), and Pro ($100/mo); per-plan generation caps apply but are not publicly disclosed. API billed per token: $5/1M text input, $8/1M image input, $30/1M image output (batch at roughly half rate). Per-image API cost ranges from ~$0.006 (low, 1024×1024) to ~$0.211 (max quality).

Key features

  • GPT Image 2.5 Flare (fast) and Sunburst (high-detail) API models with five quality tiers including new xhigh and max
  • Up to 4K (3840×2160) resolution output with wide aspect ratio support (3:1 to 1:3)
  • Thinking mode: reasons and plans before generating (available on paid plans)
  • Sketch, Templates, mobile editing/comments, and prompt sharing — no extra subscription required
  • SynthID invisible watermark and C2PA metadata on every generated image

Quick answers

What is ChatGPT Images?
OpenAI's ChatGPT Images (now powered by GPT Image 2.5, released September 8 2026) is the default image engine across all ChatGPT tiers, replacing both DALL-E 3 and the DALL-E GPT (retired August 30 2026). It generates and precisely edits images from natural language, with up to 4K resolution via the API, a thinking mode for reasoned generation, and new workflow tools including Templates, Sketch, and prompt sharing. Two API variants — Flare (speed-optimized) and Sunburst (detail-optimized) — give developers finer control over cost and fidelity.
Why does ChatGPT Images matter?
ChatGPT Images 2.5 consolidates image generation, editing, and workflow tooling into a surface that already has hundreds of millions of users, displacing standalone tools for a large share of everyday creative tasks. The retirement of the DALL-E GPT and the shift to all-tier access (with undisclosed caps) signals OpenAI treating image gen as a retention feature rather than a upsell lever — worth watching as caps tighten or plans reprice. For API buyers, the Flare/Sunburst split and five quality tiers introduce real cost-optimization handles that earlier GPT Image generations lacked. Lock-in risk is real: conversational edit history and prompt-sharing links are ChatGPT-native and do not export cleanly.
What is ChatGPT Images best for?
ChatGPT users and developers who need current-generation image generation and precise editing — from quick consumer drafts to API-scale multi-image pipelines — without leaving the chat or managing a separate image tool
How much does ChatGPT Images cost?
Image generation available on all ChatGPT plans including Free ($0), Go ($8/mo), Plus ($20/mo), and Pro ($100/mo); per-plan generation caps apply but are not publicly disclosed. API billed per token: $5/1M text input, $8/1M image input, $30/1M image output (batch at roughly half rate). Per-image API cost ranges from ~$0.006 (low, 1024×1024) to ~$0.211 (max quality).

New tools drop every week.

ONE BRIEFING · EVERY FRIDAY · FREE

Free. Unsubscribe anytime.