OpenAI launched ChatGPT Images 2.6 on 4 November, with GIFs, reference images and two API models, GPT-Image-2.6 Flare and Sunburst. Its launch post says “Images 2.6 generates images 50% faster than before.” Its API pricing page lists both new models at $30.00 per million image output tokens and $8.00 per million image input tokens, the same rates it lists for GPT Image 2.5 and gpt-image-2.
The rate per token is unchanged; what one image costs depends on how many tokens it uses. OpenAI’s own cost calculator puts one 1024×1024 image from GPT Image 2.6 at low quality at 823 output tokens, or $0.02469. Its table for gpt-image-2 lists $0.006 at low quality and the same size, which at $30 per million is 200 tokens. By our arithmetic that is about 4.1 times the tokens and the cost per image, on two settings that share a name but not a model.
Business Pill · Gross Margin Explained
An evergreen Business Engineer explainer on what an AI company keeps from each dollar. It teaches the general idea only and does not discuss OpenAI’s image models.
The key insight: As we read OpenAI’s pages, Images 2.6 changes speed and features but not the posted rate: both API models list $30.00 per million image output tokens, the same as GPT Image 2.5 and gpt-image-2, while the calculator’s default low-quality 1024×1024 image uses 823 tokens, by our arithmetic about 4.1 times the 200 implied for gpt-image-2 at its low setting.
OpenAI posted on X at 23:35 UTC that Images 2.6 is “Now with GIFs and reference images” and that “two new models are coming to the API with 50% faster image generation.” Five seconds later it posted that “GPT-Image-2.6 Flare and Sunburst are also available in the API.” The launch post says Images 2.6 is available globally in ChatGPT and Codex across all subscription tiers on desktop, mobile and web.
The post splits the two models by job: “When quality is set to low, Flare can generate images in as little as six seconds.” Sunburst, it says, brings higher-quality generation and greater image consistency across edits when quality is set to medium or high. Image markup and SVG exports come to the API, along with a new complexity parameter that sets how much planning the model does.
The post also gives a usage figure: “People around the world now create 4 billion images per week across ChatGPT, Codex, and the API.” It names one API user: “Town, a startup building an AI work assistant, used Images 2.6 in the API to explore how animated characters could bring more personality to everyday interactions at work.”
OpenAI API image output prices per million image tokens, Standard and Batch tabs of OpenAI’s pricing page as read on 4 November 2026. GPT Image 2.5 and 2.6 Flare and Sunburst share the rates shown; the Batch tab had no 2.6 row when read.
OpenAI’s launch post at 23:35 UTC on 4 November. It introduces ChatGPT Images 2.6 with GIFs and reference images and says two new API models bring 50% faster image generation; the piece quotes this line exactly and keeps the speed claim separate from the prices.
Images 2.6 is available today in ChatGPT, and Codex across all tiers on web, desktop, and mobile.
GPT-Image-2.6 Flare and Sunburst are also available in the API.
OpenAI’s availability post five seconds later. It says Images 2.6 is in ChatGPT and Codex across all tiers and that GPT-Image-2.6 Flare and Sunburst are available in the API, linking the launch post the piece uses as its primary.
What the Price List Says
On the Standard tab of the pricing page, gpt-image-2.6-sunburst and gpt-image-2.6-flare each list $8.00 input, $30.00 output and $2.00 cached input per million image tokens, plus $5.00 per million text input tokens. The same four figures appear for gpt-image-2.5-sunburst and gpt-image-2.5-flare. In the page’s all-models view, gpt-image-2 lists the same $8, $30 and $2 image-token rates, gpt-image-1.5 lists $32 output and gpt-image-1-mini $8.
The Batch tab lists gpt-image-2 and both 2.5 models at $4.00 input and $15.00 output, half the Standard rates. As we read the page at about 23:50 UTC, it had no row for either 2.6 model in the Batch tab.
OpenAI’s image generation guide warns against reading the flat rate as a flat price: “Equal token rates don’t mean equal cost per image: token consumption can differ by model and quality setting.” Its calculator note adds that “The models can use different token counts for the same quality setting and share the same price per image output token.” The guide says markup-guided edits carry no extra fee.
The Per-Image Arithmetic
The calculator on OpenAI’s guide opens on GPT Image 2.6 at low quality and 1024×1024, and shows 823 output tokens and an estimated $0.02469 per image. The guide’s historical table lists gpt-image-2 at $0.006 for the same quality name and size. Dividing $0.006 by $30 per million gives 200 tokens, our arithmetic; 823 over 200 is about 4.1.
The two are not like for like. GPT Image 2.6 has five quality settings, from low to max; gpt-image-2 has three, and the guide says the models can use different token counts for the same quality setting. The calculator figure also excludes text and image input tokens. What the arithmetic shows is narrower: at the same posted rate, the cheapest named setting on the new model uses more output tokens than the cheapest named setting on gpt-image-2.
The usage figure sets the scale. By our arithmetic, 4 billion images a week is about 571 million a day; OpenAI’s post says “more than half a billion images created per day.”
As we read the pricing page, the unit OpenAI prices has stayed the token while the unit a buyer budgets is the image; the guide itself says “Equal token rates don’t mean equal cost per image: token consumption can differ by model and quality setting.”
As we read the launch post, the two models split one menu into a speed tier and a fidelity tier at a single rate: it describes Flare as able to generate images in as little as six seconds at low quality, and Sunburst as the choice for higher-quality generation at medium or high.
OpenAI, Introducing Images 2.6, 4 November 2026
“People around the world now create 4 billion images per week across ChatGPT, Codex, and the API.”
Three Implications
FOR DEVELOPERS OpenAI’s calculator shows 823 output tokens and $0.02469 for a low-quality 1024×1024 image on GPT Image 2.6; its table lists $0.006 for gpt-image-2 at low quality and that size, a gap of about 4.1 times by our arithmetic on settings that differ by model.
FOR BATCH WORKLOADS The Batch tab lists gpt-image-2 and the GPT Image 2.5 models at $15.00 per million output tokens, half the Standard rate; when we read it, the tab had no row for the 2.6 models.
FOR CHATGPT OpenAI says Images 2.6 is available in ChatGPT and Codex across all subscription tiers, and that people create 4 billion images per week across ChatGPT, Codex and the API.
The framework puts it this way: “The new job of finance is to manage the ratio between the two — work produced per token burned.”
As we read it, an image model priced per token makes that ratio the buyer’s variable: OpenAI held the rate at $30 per million output tokens, and the tokens one image uses depend on the model and the setting chosen. The framework does not discuss OpenAI’s image models or this launch.
OpenAI’s post does not say which earlier model the 50% speed gain is measured against. Its first X post says the API models are “coming”; its post five seconds later and the launch post say they are available. The 823-token figure is the calculator’s default case, one size at one setting, and excludes input tokens. The 4.1 times ratio is our arithmetic across two models whose quality settings differ.
The Batch tab had no 2.6 row when we read it, and the pages we read do not say whether one is planned.
This Week In The AI Supercycle
Anthropic’s Reality Check, The Capital Squeeze, & The New World Order
This week’s issue: for two years the AI build-out was priced as a growthstory. This week it started being priced as a credit story.
OpenAI says Images 2.6 is 50% faster and lists its two API models at the same $30 per million output tokens as GPT Image 2.5 and gpt-image-2. Its own calculator puts a low-quality 1024×1024 image at 823 tokens, about 4.1 times the 200 implied for gpt-image-2 at its low setting, our arithmetic.
A note on sourcing. Prices are from OpenAI’s API pricing page, read at about 23:50 UTC on 4 November 2026, and per-image figures from OpenAI’s image generation guide and its calculator, read the same day; product claims are from OpenAI’s launch post and its posts on X. The 200-token figure, the 4.1 times ratio and the per-day image count are our arithmetic.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
