- OpenAI’s Codex can now operate autonomously on macOS desktops — clicking, typing, and running applications in the background without constant supervision.
- A built-in browser, 90+ integrated plugins (Atlassian, Microsoft Suite), and native image generation via gpt-image-1.5 extend Codex far beyond coding into full workflow automation.
- A new pay-as-you-go pricing tier for ChatGPT Enterprise and Business customers signals OpenAI’s push to capture corporate budgets currently flowing to Anthropic’s Claude Code.
- Sam Altman took a veiled swipe at Anthropic hours after the Claude Opus 4.7 launch, criticizing rivals for “reduced limits or weaker models” — escalating the most consequential rivalry in enterprise AI.
OpenAI this week unveiled a sweeping overhaul of Codex, its AI-powered coding tool, transforming it from a developer-focused assistant into what the company now positions as a universal work automation agent. The update arrives at a strategically charged moment — just hours after Anthropic launched Claude Opus 4.7 — and represents OpenAI’s most aggressive move yet to reclaim ground in the enterprise AI tools market where Anthropic’s Claude Code has been steadily gaining share.
From Coding Assistant to Desktop Agent
Autonomous Computer Control
The centerpiece of the update is full desktop autonomy. Codex can now “operate in the background on your computer — opening any app on your desktop and carrying out operations with a cursor that clicks and types,” according to TechCrunch’s report. On macOS, this means Codex can independently navigate applications, execute multi-step workflows, and even schedule future tasks that it will autonomously wake itself to complete — functioning less like a chatbot and more like a digital colleague.
Browser, Plugins, and Creative Tools
OpenAI also shipped an in-app browser that lets Codex execute commands on web applications — initially for front-end and game development, with plans to expand to full browser control. The plugin ecosystem now includes over 90 integrations spanning Atlassian, Microsoft Suite, Slack, and Google Calendar, letting Codex handle clerical work like compiling to-do lists from your Slack channels and calendar. Perhaps most surprising is the addition of native image generation through gpt-image-1.5, enabling front-end layout design and gaming asset creation directly within the coding environment.
The Anthropic Factor
Strategic Timing and Market Dynamics
The timing of the Codex overhaul was unmistakably deliberate. OpenAI launched the update just hours after Anthropic announced Claude Opus 4.7 with its breakthrough 98.5% visual accuracy and 13% coding benchmark improvement. Sam Altman added fuel to the fire with veiled remarks criticizing rivals for “reduced limits or weaker models” — widely interpreted as a direct shot at Anthropic CEO Dario Amodei. The two leaders famously refused to share a stage at the India AI Impact Summit earlier this year.
The competitive stakes are enormous. Anthropic’s Claude Code has reached an annualized revenue run rate of $2.5 billion in just nine months, while close to a third of American businesses now pay for Anthropic’s AI tools — a figure that jumped more than six percentage points in a single month. OpenAI’s response is to widen the surface area of competition: rather than matching Claude Code feature-for-feature on coding benchmarks, Codex is evolving into a broader automation platform that bundles coding, creativity, and workflow management into a single agent.
Enterprise Pricing Play
The new pay-as-you-go pricing model for ChatGPT Enterprise and Business customers is a clear signal that OpenAI is targeting corporate procurement budgets. The $100-per-month ChatGPT Pro subscription tier offers five times the Codex usage of the $20 Plus plan, while enterprise customers gain flexible consumption-based billing. With over 3 million active Codex users, OpenAI is betting that platform breadth — not just model performance — will be the deciding factor in enterprise adoption.
AI Biz Insider Analysis ― OpenAI is making a classic platform play: if you cannot beat the specialist (Claude Code on coding), change the game by bundling more capabilities into a single agent. The risk is that breadth introduces complexity and dilutes the core coding experience — precisely the domain where Anthropic’s focus has earned it developer loyalty. The real winner may be enterprise buyers, who now have genuine leverage to negotiate between two fierce competitors. Watch for Anthropic’s counter-move; the company has been quietly shipping Claude Code integrations with VS Code, JetBrains, and terminal environments that deepen developer lock-in through workflow muscle memory rather than feature checklists.
Related
- Cloudflare Absorbs Replicate and Builds a Unified Inference Layer for the Agentic Era
- Manycore Tech Surges 185% on Hong Kong Debut — First of Hangzhou’s ‘Little Dragons’ Goes Public
- Adobe Just Turned Its Entire Creative Suite Into One AI Agent
- Factory AI’s ‘Droids’ Hit $1.5B Valuation as Enterprise Coding Agents Boom
Sources
- TechCrunch — OpenAI takes aim at Anthropic with beefed-up Codex (April 16, 2026)
- Republic World — OpenAI Supercharges Codex Into ‘AI Super App’, Sam Altman Takes Swipe at Anthropic (April 17, 2026)
AI Biz Insider · AI Business EN · aibizinsider.com
댓글 남기기