Overview
On April 16, 2026, OpenAI shipped a major update to Codex internally referred to as "Codex for (almost) everything." Unlike the February 5 GPT-5.3-Codex release, which was a model upgrade, this update is primarily an app-surface expansion around Codex rather than a benchmark-focused model launch, adding new capabilities across the work surface, integrations, and continuity.
The update positions Codex less as an isolated coding assistant and more as a workflow control surface for the 3M+ developers who already use it weekly. New capabilities include background computer use on macOS, an in-app browser with inline page comments, native image generation via gpt-image-1.5, 90+ plugins combining skills and MCP servers, GitHub PR review built into the app, multi-terminal tabs, SSH access to remote devboxes (alpha), and a memory preview system. Availability is uneven — several features roll out macOS-first, with EU/UK, Enterprise, and Education tenants receiving them on a staggered schedule.
What's New
Background Computer Use on macOS
Codex can now operate other applications on your Mac by seeing the screen, moving its own cursor, and typing — while you keep working in parallel. Multiple agents can run simultaneously without hijacking your active window or input focus. This expands Codex from a coding-focused assistant into a broader desktop workflow tool, bringing AI agent computer-use capabilities into the live macOS app.
In-App Browser With Inline Comments
Codex now includes an early in-app browser for local or public pages that do not require sign-in, letting you comment directly on rendered pages while iterating on frontend designs, apps, and games. You can comment directly on rendered pages to give Codex precise feedback, which shortens the ideate → render → review → iterate loop inside one workspace.
Image Generation via gpt-image-1.5
Codex integrates gpt-image-1.5 for in-app image generation and editing. Use cases OpenAI highlighted include UI mocks, placeholder asset creation, and visual checks during frontend iteration. OpenAI positions the feature for UI mocks, placeholder assets, and visual iteration inside the same workflow.
90+ New Plugins
Codex gains 90+ new plugins that bundle skills, app integrations, and MCP servers together. Named launch partners include Atlassian Rovo, CircleCI, CodeRabbit, GitLab Issues, Microsoft Suite, and Render. Each plugin combines skills, app integrations, and MCP server configuration so Codex can gather context and take actions across connected tools.
Developer Workflow Upgrades
Several engineering-team features landed in the same release:
- GitHub pull request review inside the Codex app, including sidebar inspection, diff review, and support for addressing review comments
- Multiple terminal tabs for parallel command sessions within one agent
- SSH access to remote devboxes (alpha) for running Codex against long-lived cloud environments
- Richer file previews and a new summary pane for faster context scanning
Memory Preview & Continuity
A memory preview surface now stores user preferences, prior edits, and context gathered during past tasks. Codex can resume threads and schedule future work, making it practical to hand off multi-hour or cross-session workflows — issue triage, CI failure summaries, release briefs, and recurring bug checks are called out as first-party automations.
Availability & Access
Rollout is uneven and worth checking before relying on a specific feature:
- Background computer use launches on macOS first; EU and UK availability is scheduled but delayed.
- Memory and context-aware suggestions roll out gradually for Enterprise, Education, EU, and UK tenants.
- Remote devboxes via SSH remain in alpha.
- Windows support for the desktop app exists (added March 4, 2026) but the April expansion leans heavily on macOS-specific computer-use APIs.
OpenAI frames this release as a Codex app update rather than a new Codex model launch.
Pricing & Plans
OpenAI has not announced pricing changes tied to this release. Codex access continues to follow the existing tiers:
- ChatGPT Plus ($20/month): Access to Codex with standard rate limits
- ChatGPT Pro (from $100/month): Pro currently has $100 and $200 tiers, both with higher Codex usage than Plus.
- Business / Enterprise / Edu: Business includes self-serve standard seats and optional usage-based Codex seats; Enterprise and Edu are admin-managed offerings with staged rollout for some personalization features.
- API: Codex capabilities available via OpenAI's developer API with usage-based billing
Usage and billing vary by plan and, for some workspaces, by credits or flexible pricing; check the current Codex pricing and plan documentation for the latest limits.
Best For
- Engineering teams already running Codex on macOS who want to replace ad-hoc automation scripts with in-app workflows.
- Frontend developers iterating on web UI who can benefit from the browser + image-gen loop.
- DevOps and platform teams adopting plugins for CircleCI, Render, GitLab, or CodeRabbit alongside Codex.
- Individual developers on ChatGPT Plus looking to test agentic desktop workflows without upgrading plans.
- Teams using Codex for long-running background tasks (issue triage, CI summaries, release briefs) that benefit from memory and thread resumption.
FAQ
Is "Everywhere" a new Codex model?
No. The underlying model remains GPT-5.3-Codex, which shipped on February 5, 2026. The April 16 update is a product and platform expansion around that model.
Do I need to upgrade my ChatGPT plan to get these features?
No new paid upgrade is required for eligible desktop-app users, but access depends on plan and rollout. Codex is included with Plus, Pro, Business, and Enterprise/Edu, and for a limited time also with Free and Go; some features still roll out separately by region and workspace type.
Does background computer use work on Windows?
Not at launch. The background computer use capability rolled out on macOS first. Windows users can continue using the Codex desktop app (shipped March 4, 2026) but without the parallel desktop-agent behavior.
What happens to my data with the new memory features?
Memory stores preferences, edit history, and gathered context across threads. Enterprise and Education tenants see memory roll out gradually with admin controls. Review your organization's data-retention settings before enabling memory on sensitive workloads.
Can I still use Codex over SSH on remote servers?
Yes, but SSH access to remote devboxes is currently in alpha. Expect breaking changes and limited support until the feature moves to GA.
Compatibility Notes
- Prior Codex extensions and CLI workflows continue to work; plugins are additive.
- GPT-5.4 (separately announced for long-context work) is not the engine behind this update — teams needing 1M-token context should evaluate GPT-5.4 independently of this release.
