San Francisco, CA — OpenAI is rolling out flexible pricing for Codex, giving teams a new way to scale AI-powered software development without being locked into rigid usage limits. The move is designed to make Codex more accessible for growing engineering teams and enterprise workflows.
Instead of fixed quotas, teams can now purchase additional usage on demand, turning Codex into a more scalable, consumption-based AI development platform.
From Fixed Limits to On-Demand Scaling
Previously, Codex usage was largely tied to subscription tiers like Plus, Pro, Business, or Enterprise. Now, OpenAI is introducing a flexible, credit-based model layered on top of these plans.
This means:
Teams can buy extra credits when needed
No need to upgrade entire plans for short-term spikes
Usage scales with actual development workload
👉 In simple terms: pay for what you use, when you need it
Built for Real Engineering Teams
The update is clearly targeted at organizations using Codex as a core part of their development workflow.
With flexible pricing, teams can:
Run large-scale code generation tasks
Execute multiple AI agents simultaneously
Handle peak workloads (releases, migrations, refactors)
Avoid hitting hard usage caps mid-project
This aligns with Codex’s evolution into a full software engineering agent, not just a coding assistant.
How Pricing Works Across Plans
Codex is still bundled inside ChatGPT plans, but now includes more flexibility:
Plus / Pro users → get baseline usage with rate limits
Business plans (~$25/user/month) → team collaboration + admin controls
Enterprise → custom pricing with security and compliance features
On top of this:
👉 Teams can purchase additional credits instead of upgrading plans
This hybrid model combines:
Subscription predictability
Pay-as-you-go flexibility
Codex pricing reflects a broader shift:
👉 From fixed SaaS subscriptions → to usage-based AI infrastructure
We’re seeing AI tools behave more like:
Cloud services (pay for compute)
APIs (pay per request)
Agent platforms (pay per task executed)
Source: OpenAI
Join the conversation! Your thoughts help the community grow.