SShortSingh.
TopTechnologyAI & MLStartupsProgrammingScienceBusinessFinanceCrypto & Web3World
Back to feed
Programming

OpenAI May Be Building a Sketch Tool for ChatGPT's Image Workflow

0
·6 views

OpenAI appears to be developing a sketch-based input feature for ChatGPT that would allow users to draw rough visuals directly in the chat interface instead of relying solely on text prompts. A credible September 2026 Axios report cited OpenAI product lead Adele Li discussing a sketch tool and templates in connection with the upcoming ChatGPT Images 2.5 update. The feature has not been officially confirmed or documented by OpenAI, and details such as naming, availability, and release timing remain unspecified. OpenAI has, however, confirmed two prior image milestones: a refreshed ChatGPT Images experience in December 2025 and the launch of ChatGPT Images 2.0 in April 2026, which introduced improved editing and a new creation space. A sketch tool would logically extend this multimodal direction by letting users reduce prompt ambiguity through simple visual references rather than lengthy written descriptions.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Log in to join the discussion and vote.

Log in

Related stories

0
ProgrammingDEV Community ·

Oracle OCI AI Foundations Course Breaks Down AI, ML, and Deep Learning Basics

Oracle University offers an OCI AI Foundations course that introduces learners to artificial intelligence, machine learning, deep learning, neural networks, and large language models. The course distinguishes between these often-confused terms, explaining that AI is the broadest field, machine learning is a data-driven subset of AI, and deep learning is a further specialization using multi-layered neural networks. Rather than focusing solely on generative AI, the course builds foundational knowledge before connecting concepts to Oracle's cloud and database ecosystem. It uses practical examples, such as predicting employee turnover, to illustrate how machine learning models identify patterns from historical data instead of relying on manually coded rules. The training is designed for professionals working with Oracle technologies who want to understand how modern AI capabilities integrate with enterprise applications.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Dev Reflects on Building a Secure Inventory Router for a Restaurant Management App

A software developer has shared a detailed retrospective on their contribution to Dine Flow, a restaurant inventory and order-management platform built by a six-person team over six weeks. Their core responsibility was the Express/TypeScript backend, with a particular focus on the stock router powered by Prisma and PostgreSQL. The router handled ingredient tracking, expiration dates, supplier integrations, and automatic stock decrements tied to order placements. During development, the engineer caught a subtle bug where a single-item endpoint was silently returning the entire stock table with a 200 status, reinforcing the lesson that a successful response and a correct response are not the same thing. A deliberate design choice was also made to treat each delivery as an immutable record rather than updating existing rows, preserving batch-level expiration tracking critical to food safety.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Open-Weight Models vs. Open Stack: Understanding the Difference in AI Transparency

Open-weight AI models, such as Llama, Mistral, and Gemma, allow users to download and run model weights on their own infrastructure, but they typically share little about how the model was built. An 'open stack' goes further by making the full development pipeline more transparent, including training data, code, recipes, evaluation methods, and deployment tools. Openness in AI is not a fixed standard but rather a spectrum, with some projects releasing model weights and architecture while keeping training and evaluation details private. Projects like OLMo by Ai2 and BLOOM by BigScience represent more comprehensive open approaches, sharing not just the final model but the entire process behind it. Understanding this distinction helps developers and researchers assess how much insight they truly have into the AI tools they use.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

OpenAI admits it cannot rule out training on mathematician's unpublished Navier-Stokes work

NYU math professor Tristan Buckmaster and collaborator Levent Alpöge spent nearly a year working toward a proof of finite-time blowup for fluid equations — a problem related to the Navier-Stokes Millennium Prize — feeding drafts into OpenAI's Codex throughout. In early September, Buckmaster learned that an internal OpenAI model had independently produced a similar proof, using the same smooth-forcing setup he and Alpöge had quietly chosen. He directly asked OpenAI whether his Codex sessions had been used in training, but received no clear answer until the company issued a public statement. OpenAI's statement confirmed that while no user data was directly 'accessed' or 'seen' by its model, it 'cannot rule out' that de-identified data from their product usage helped improve its models. The episode highlights that consumer Codex sessions are used as training data by default, raising broader questions about the use of unpublished professional work submitted through AI coding tools.

0 comments Read more at DEV Community