Comparable rates
Use the same official rate card as the starting point for both models.
Developer cost guide
GPT Image 2.5 API Pricing
See the official GPT Image 2.5 Flare and Sunburst token rates, understand what changes real image cost, and test the workflow with visible site credits.
$5 / 1M text input tokens$8 / 1M image input tokens$30 / 1M image output tokens
Open the workflow
Sign in is required before generation. The selected model and credit cost stay visible.
Workflow example
An image API workflow showing inputs, model processing, review, and delivery
Controls that match the task
Only the inputs that affect this workflow are shown.
Clear before you run
Model, availability, limits, and paid cost stay visible.
A usable result
Preview the output, then download or continue editing.
Verified workflow
Use the model, not a mock demo
The workspace submits to the named GPT Image 2.5 endpoint. It does not silently replace your selection with another model.
Official rate snapshot · September 29, 2026
Text input
$5 / 1M
Cached text
$1.25 / 1M
Image input
$8 / 1M
Cached image
$2 / 1M
Image output
$30 / 1M
Both official model pages list these rates. They are not an exact per-image quote.
Step 01
Configure the input
GPT Image 2.5 Flare
Official provider identity
155/32,000
This page always submits to GPT Image 2.5 Flare. It never substitutes another model. Availability and billing are verified by the server before a task starts.
Step 02
Review the result
Example · site-owned workflow illustration
High-detail campaign concepts. This illustration is not presented as a live output from this model.
Recipe: A refined launch campaign key visual with controlled typography-safe negative space.
mode: generate · ratio: 16:9 · quality: high
Verified against primary sources
Positioning, inputs, quality tiers, endpoints, and prices were checked against the official model pages.
Designed around the result
Estimate the whole workflow, not just one token rate
The official pages currently list matching rates for Flare and Sunburst. Actual spend changes with token use, input images, quality, retries, and candidate count.
Task-level budgeting
Measure prompt, reference input, output, retries, storage, and delivery separately.
Visible site billing
This generator shows its credit quote before submission; credits are not API tokens.
A predictable workflow
From input to finished file
Each stage answers a practical question before the next action is available, so you can spot a wrong input before spending time or credits.
- STEP 1
Choose the workflow
Separate generation, editing, exploration, and final rendering.
- STEP 2
Fix the quality policy
Decide when low or medium is enough and when xhigh or max is justified.
- STEP 3
Measure accepted outputs
Track cost per approved asset, not only cost per API request.
- STEP 4
Add operational costs
Include moderation, storage, failed jobs, support, and bandwidth.
Get a cleaner first result
Date every pricing snapshot.
Keep API tokens separate from site credits.
Budget retries by task type.
Use lower quality for exploration and maximum quality selectively.
Practical answers
Questions before you start
What are the current GPT Image 2.5 API token rates?+
As checked on September 29, 2026, both official pages list $5 per million text input tokens, $1.25 cached text input, $8 image input, $2 cached image input, and $30 image output.
Is Sunburst more expensive than Flare?+
The official pages currently list the same token rates. Total task cost can still differ because token use and retry behavior differ.
Can I calculate an exact image price from the rate card?+
Not from headline rates alone. You need actual token use, quality, inputs, output, and retries. OpenAI notes its GPT Image 2 calculator does not estimate GPT Image 2.5 token use.
Are QwenImage3 credits the same as API tokens?+
No. Site credits are the billing unit shown before a run; API tokens are the provider’s underlying unit.
Keep creating
Useful next steps
Ready when you are