What changed on 4 November?
OpenAI announced Images 2.6 for ChatGPT and Codex globally across subscription tiers on desktop, mobile and web. It also introduced Flare and Sunburst API variants, GIF workflows, image markup in the API, SVG exports and a complexity control separate from image quality.
Tier availability does not establish an unlimited free allowance. OpenAI’s claims about speed and output quality are vendor claims; Fewertools has not run a generation or reproduced a benchmark. Read the launch announcement.
API usage has its own billing basis
The Standard image-generation pricing table lists the same rates for GPT-Image-2.6 Flare and Sunburst:
| Token category | Input | Cached input | Output |
|---|---|---|---|
| Image | $8 | $2 | $30 |
| Text | $5 | $1.25 | Not quoted |
These are token rates, not a price per image, a monthly subscription or a total job quote. Input images, text and generated output have different rates. The total requires the actual token quantities; no Batch discount or cache eligibility for these variants is assumed here.
Keep the product and API bills separate
A ChatGPT plan and an API usage bill answer different cost questions. The ChatGPT profile keeps its dated consumer-plan observations. The GPT-image-2 profile remains a historical model entry, with a pointer to this newer release rather than silently relabelling its earlier evidence.
Use Check my subscriptions for supported recurring plans, and the stack calculator for your separately entered API bill. Neither determines a token-priced invoice from this announcement. Source receipts and limitations.