Superpower Daily
Google Signs Canada’s Data-Centre Principles, Pledges to Cover...GPT-6 Astra Hits 99.9% on ARC-AGI-3, but Scores 62.7% in a Shar...Crusoe Signs Reported $13 Billion, Five-Year AI Cloud Deal With...Microsoft Releases MAI-Transcribe-2 at 10 Cents an HourxAI Opens Grok Bot to Enterprises With Controls for Autonomous...Meta Lawsuit Adds Bystanders Who Say AI Glasses Recorded Them W...Google Signs Canada’s Data-Centre Principles, Pledges to Cover...GPT-6 Astra Hits 99.9% on ARC-AGI-3, but Scores 62.7% in a Shar...Crusoe Signs Reported $13 Billion, Five-Year AI Cloud Deal With...Microsoft Releases MAI-Transcribe-2 at 10 Cents an HourxAI Opens Grok Bot to Enterprises With Controls for Autonomous...Meta Lawsuit Adds Bystanders Who Say AI Glasses Recorded Them W...
Productspublished

OpenAI’s Codex Shows a Gated Image Upgrade, With Access Still Unclear

The hidden interface offers customer-facing promises around image quality, speed and creative controls, while leaving the underlying technology and release terms unspecified.

By 2 min read
OpenAI’s Codex Shows a Gated Image Upgrade, With Access Still Unclear
OpenAI’s Codex Shows a Gated Image Upgrade, With Access Still Unclear

Listen to this story

The audio brief

About 1:28
0:001:28
Read transcript
A hidden modal inside OpenAI’s Codex is promising a major upgrade to image creation, but the feature is still locked behind a disabled gate. The message appeared on September third with a “Try it” button, describing higher-quality results, faster generation, and smarter creative tools. That is meaningful product staging, but it is not a launch. The interface suggests OpenAI has prepared customer-facing onboarding for a new image experience inside Codex. It does not say when the feature will activate, who will get access, or whether anyone can use it yet. The clearest promise is speed. Shorter generation times could make repeated edits more practical, especially when users are refining composition, changing text, or adapting an image to a different format. But the modal does not identify the underlying technology. There is no model name, benchmark, output resolution, pricing change, API access, or rollout plan. “ImageGen 2.5,” a label circulating outside the interface, is not an official name shown here. For now, the documented public option remains ChatGPT Images 2.0, available across subscription tiers. That follows OpenAI’s retirement of the official DALL-E GPT on August thirtieth, with users directed instead to ChatGPT Images. The key thing to watch is whether this gated promise becomes a selectable product—and what technical and access details OpenAI discloses when it does.

Story brief

3 key points

OpenAI appears to be preparing a next-step image-creation experience inside Codex, but the feature remains disabled and cannot be treated as a launch. A September 3 modal promises faster generation and improved creative tooling without identifying a model, benchmark, resolution, price, API, or release date. ChatGPT Images 2.0, released April 21, remains the documented public option. The practical signal is product...

  1. 01

    The modal appeared September 3 with “Try it,” but its disabled feature gate leaves activation timing and eligible users unknown.

  2. 02

    “ImageGen 2.5” is an external label; OpenAI’s interface does not name a successor model or disclose technical specifications.

  3. 03

    Faster generation could make iterative edits to composition, text, and formats more practical.

A hidden modal inside Codex points to a prepared upgrade for ChatGPT image creation. It promises higher-quality results, faster generation and smarter creative tools, but the disabled feature gate means the interface does not show public customer access.

A customer message waiting behind a gate

The modal surfaced on September 3 with the headline “Image creation got a major upgrade” and a “Try it” button. The person who surfaced the screenshot said the feature had already been built into the product but was held behind a disabled feature gate.

That mechanism matters: a feature gate is a product control that can keep built software inactive. Here, it supports a narrower conclusion than a launch announcement—the upgrade has customer-facing onboarding copy inside Codex, while the timing and audience for activation remain unknown.

The promises describe an experience, not the engine

Faster generation is the most concrete promise in the modal. Shorter waits would make repeated image revisions easier, especially where a user needs to change composition, text or format rather than accept a first output.

“Smarter creative tools” leaves the scope open. The visible copy cannot distinguish a new underlying model from changes to the system serving it, added editing controls, or a package combining those elements.

The material unknowns

  • No model architecture, benchmark or output resolution is disclosed.
  • No pricing structure, API endpoint, activation date or rollout plan is disclosed.
  • “ImageGen 2.5” is an externally supplied label, not a name shown in the interface.

The current public product remains Images 2.0

OpenAI’s public documentation still identifies ChatGPT Images 2.0 as the available image-generation model across subscription tiers. OpenAI released Images 2.0 on April 21, so the hidden modal appears to concern a prospective successor or product update rather than a model users can select today.

The staging comes days after OpenAI retired the official DALL-E GPT from ChatGPT on August 30 and directed users to ChatGPT Images. The result is a simpler product boundary: ChatGPT Images is the public destination, while the next capability step remains defined only by the gated modal’s broad promises.

Related coverage

Continue reading