JAPANITIESJapan and France, through words and food.
[Thursday 5 November] Strike and demonstrations. Bordeaux: meet at 12:30 pm at place de la République, 4 km to place de la Victoire. See the details
Tech newsJAPANITIES

OpenAI launches “Images 2.6”: 50% faster image generation and animated GIFs

Checked on

Published on

AT A GLANCE
  • On 4 November OpenAI presented “Images 2.6”: generation is 50% faster and it can also make animated GIFs.
  • Available in every ChatGPT and Codex plan, free included, worldwide; GIFs are rolling out gradually.
  • Two API models (Flare, Sunburst), prices unchanged from Images 2.5, with SVG export added.
  • The speed figures are the company’s own claims, not measured independently.
Illustration photo of a person holding the OpenAI icon. Not a screenshot of Images 2.6.
FoxTPNL · CC BY 4.0Illustration photo of a person holding the OpenAI icon. Not a screenshot of Images 2.6.
In this article

The announcement

According to ITmedia NEWS (5 November, 10:59), OpenAI announced “ChatGPT Images 2.6” on 4 November (US time), the successor to Images 2.5, launched in September. Image generation is 50% faster and editing is more accurate. It is available from 4 November in every ChatGPT and Codex plan, including the free one, worldwide (web, desktop, mobile). GIF generation is being rolled out gradually.

New features

(1) Animated GIFs: you can ask for looping animations, or upload a hand-drawn sketch to make a character move, for example wave; the speed can be adjusted and the GIF shared. The uses mentioned are company mascots, product animations and reaction images for social networks. (2) Saved reference images: you name a photo (the example is a dog called “Pepper”) and then reuse it in other conversations with a tag such as “@Pepper”, without uploading it again. (3) Editing consistency: if you change a character’s background, the costume and details are kept.

API for developers and pricing

The API offers two models. GPT-Image-2.6 Flare favours speed: at the low-quality setting it can generate an image in about 6 seconds. GPT-Image-2.6 Sunburst aims at medium to high quality, with more detail and more consistency when editing. The API also adds annotation of areas of an image to indicate changes, SVG export and a “complexity” parameter that sets how long the model thinks before generating. Prices are the same as Images 2.5, per million tokens and for both models: image input 8 dollars, image output 30 dollars, text input 5 dollars.

Safety, scale of use and unverified points

OpenAI published a system card the same day and describes new measures to detect harmful motion in GIFs. C2PA metadata is kept and Google DeepMind’s SynthID watermark is to be added to ChatGPT, Codex and the API. According to OpenAI, about 4 billion images are created every week across ChatGPT, Codex and the API, more than 500 million a day on average. Figures such as “50% faster” or “6 seconds” are the publisher’s claims, not measured here. The exact schedule for the GIF rollout by country could not be verified.

LANGUAGE NOTES

Learn the grammar

Explained in English

French grammar

Learn sentence patterns and usage from the article’s own sentences.

01Negation and its scope
Structure

ne / n’ … pas · ne … plus

In the article
Le calendrier précis du déploiement des GIF selon les pays n’a pas pu être vérifié.
Key points

Normally ne and pas wrap the verb. Before a vowel, ne becomes n’. Check which part of the sentence the negation covers. Ne … plus means “no longer”.

02Citing a source with selon
Structure

selon + noun / organisation

In the article
Selon ITmedia NEWS (5 novembre, 10 h 59), OpenAI a annoncé le 4 novembre (heure américaine) « ChatGPT Images 2.6 », successeur d’Images 2.5 lancé en septembre.
Key points

Selon means “according to”. Put a person or organisation after it as the source. It separates facts the writer checked directly from what another body or report said. Citing a source does not turn a forecast into an accomplished fact.

03Pouvoir: ability and possibility
Structure

peut / peuvent / pourrait + infinitive

In the article
(1) GIF animés : on peut demander des animations en boucle, ou envoyer un croquis dessiné à la main pour faire bouger un personnage, par exemple qui salue ; la vitesse se règle et le GIF se partage.
Key points

Put a base-form verb after pouvoir. Peut means “can” or “may”; the conditional pourrait gives a more reserved possibility, “might”. It does not necessarily mean implementation is settled.

04Purpose with pour
Structure

pour + infinitive

In the article
S’ajoutent l’annotation de zones d’une image pour indiquer les modifications, l’export en SVG et un paramètre « complexity » qui règle le temps de réflexion avant la génération.
Key points

This is the purpose, “in order to…”. Put the base form of a verb after pour. The purpose stated here is separate from whether the effect was actually achieved.

05The passé composé for events
Structure

present of avoir + past participle

In the article
OpenAI a publié le même jour une fiche système (system card) et décrit de nouvelles mesures pour détecter les mouvements nuisibles dans les GIF.
Key points

It reports past events, as in “announced” or “explained”. A / ont is the auxiliary avoir and the next word is the past participle. The passé composé alone does not tell you the date of the event.

06The present of être
Structure

est / sont + complement

In the article
La génération d’images est 50 % plus rapide et la précision de l’édition s’améliore.
Key points

Est is third-person singular and sont third-person plural of être. They express “is / are” — a state or description. If a past participle follows, it is a different structure such as the passive, so check the words that follow.

Explained in English

Japanese grammar

Understand the structure and usage from the text.

01Citing a source: によると
Structure

noun + によると

In the article
ITmedia NEWS(5日10時59分)によると、OpenAIは11月4日(米国時間)に「ChatGPT Images 2.6」を発表しました。
Key points

As in “LCPによると”, it shows where the information comes from. What that source reported follows. It is used differently from “私にとって”, which gives one’s own opinion.

02Polite negation and plain form
Structure

〜ではありません / 〜ではない

In the article
「50%速い」「6秒」などの数字は発表側の主張で、ここで独立に測ったものではありません。
Key points

They negate a noun and similar content. “ではありません” is the polite form and “ではない” the plain form. “成立ではありません” shows that it cannot be said to have been concluded.

03Giving examples with など
Structure

noun / expression + など

In the article
企業のマスコット、商品のアニメ、SNSのリアクション画像などの用途が挙げられています。
Key points

“暴言など” gives “abusive language” as one example. It is not necessarily a closed list of only the things named. It also helps when reading the scope of a news story.

04The polite past ました
Structure

verb stem (連用形) + ました

In the article
OpenAIが11月4日(米国時間)に、ChatGPTの画像生成の新版「Images 2.6」を発表しました。
Key points

In “発表しました”, ました makes a verb polite past. The negative is “ませんでした”. Read the actual date together with the date and time in the text.

05Quoting with と: content and report
Structure

plain-form sentence + と + verbe de parole (報じる, 説明する…)

In the article
GPT-Image-2.6 Sunburstは、中〜高品質向けで、細部と編集時の一貫性が改善されたとされます。
Key points

と takes the content that is reported or explained. In “改定する方針だと報じました”, “改定する方針だ” is what the report said. Separating the content from who said it and how makes long sentences easier to read.

06Possibility in Japanese: できる・られる
Structure

verb potential form / 〜ことができる

In the article
新機能、利用できる範囲、料金、安全対策を、ITmedia NEWSの報道から整理します。
Key points

できる is the basic verb for ability. The potential form of 受ける is 受けられる. The same られる can also be passive or honorific, so decide from the meaning of the sentence.

Related articles