OpenAI unveils ChatGPT Images 2.5, slashing image generation latency up to 50% versus Images 2.0 and adding Sketch drawing in ChatGPT
openai
| Source: Techmeme | Original article
OpenAI unveiled ChatGPT Images 2.5, promising up to 50% faster image generation than its predecessor and adding a Sketch tool for drawing directly within ChatGPT.
OpenAI rolled out a new version of its image‑generation engine, ChatGPT Images 2.5, on Tuesday. The company says the upgrade slashes generation latency by as much as 50 percent compared with the previous Images 2.0 model, a notable gain for a service that has long been bottlenecked by the compute‑heavy autoregressive architecture of the GPT‑Image line.
Beyond speed, Images 2.5 adds a “Sketch” tool that lets users draw directly inside the ChatGPT interface, turning rough strokes into refined visuals. The model also promises higher detail, more precise editing, and better reference fidelity, with multi‑turn edit instructions now handled more reliably. OpenAI introduced two new style presets – Flare and Sunburst – and announced that API usage will be priced at twice the per‑token rate of the earlier model.
The improvements matter because faster turnaround lowers the cost of interactive workflows and makes real‑time visual assistance more practical for developers, designers, and end‑users. Enhanced editing precision could broaden the appeal of AI‑generated imagery in marketing, product design and education, where iterative refinement is essential.
What to watch next: how developers respond to the higher API price point and whether the Sketch feature spurs new integrations within ChatGPT’s broader ecosystem. Analysts will also be tracking latency benchmarks from competing providers, as OpenAI’s claim of a 50 percent cut sets a new performance baseline for generative image models.
Sources
Back to AIPULSEN