Learn
Is Codex Free? GPT-5.6, ChatGPT Work, CLI, Credits, and API
Yes—Codex is included in ChatGPT Free. Free and Go use GPT-5.6 Terra in Work and Codex; higher tiers add model choice and usage, while API keys are billed separately.
Clarify the spend threshold before you commit. Use this page when the core product is familiar and the real question is whether to stay free, upgrade, or switch pricing tracks.
Editorial guide
Guide
Start with the spend threshold and the conditions that change the pricing decision.
Codex is free to use as a native feature within ChatGPT, but its capabilities, model tiers, and execution allowances are strictly governed by your account tier. Integrated directly into the ChatGPT web interface and native desktop applications, Codex allows developers and technical teams to run agentic coding tasks, inspect codebases, execute terminal commands, and draft automated code edits without an upfront subscription.
However, treating free Codex access as an unmetered engineering workstation is misleading. Free accounts are anchored to lightweight base models, enforce strict hourly task caps, and restrict automated CLI pipelines. Software engineers, technical leads, and founders evaluating Codex must understand how free access differs from ChatGPT Plus, ChatGPT Pro, and pay-as-you-go API infrastructure.
Codex access inside ChatGPT Free: GPT-5.6 Terra capabilities
Every registered user with a free OpenAI account can access Codex directly inside the unified ChatGPT interface.
On the free tier, Codex operates under clearly defined technical boundaries:
- Default base engine (GPT-5.6 Terra): Free Codex sessions are powered by GPT-5.6 Terra, OpenAI’s balanced, high-speed foundational model. While Terra handles syntax parsing, single-function generation, documentation lookup, and basic unit test creation adeptly, it lacks the deep autonomous reasoning capabilities of flagship models.
- No Sol, Max, or Ultra reasoning modes: Advanced capabilities—specifically GPT-5.6 Sol and multi-agent Ultra orchestration—are locked behind paid plans. Complex full-stack refactoring, cross-repository dependency tracing, and subtle architectural debugging require higher reasoning tiers.
- Hourly and daily rate limits: Free users operate under dynamic rate limiting. During periods of heavy compute demand, free Codex sessions are subject to cooldown windows or prompt queues, making continuous multi-hour coding sprints impossible without interruption.
For students, junior developers, and hobbyists building standalone scripts or learning new frameworks, free Codex provides an exceptional learning sandbox. For production codebases requiring deep multi-file awareness, free limits quickly become a bottleneck.
The unified desktop experience: Chat, Work, and Codex explained
OpenAI has consolidated its productivity surfaces into a single desktop application environment across macOS and Windows. This interface presents three distinct operational modes available on every plan:
Surface mode | Core purpose | Primary interaction model | Model access on Free plan | Model access on Plus & Pro |
|---|---|---|---|---|
Chat | General conversation, quick Q&A, brainstorming | Back-and-forth conversational dialogue | GPT-5.6 Terra with shared limits | Flagship Sol, advanced voice, unlimited standard chat |
Work | Document authoring, data analysis, research deliverables | Canvas document editor, spreadsheet synthesis | Standard compute with file attachment limits | Deep research tools, extensive multi-document context |
Codex | Autonomous coding, repository indexing, diff patches | File tree inspection, terminal execution, patch review | GPT-5.6 Terra, basic single-file diffing | GPT-5.6 Sol, Max reasoning, automated test execution |
Across free and lower-tier accounts, usage is pooled: running extensive tasks in Codex counts against your overarching account activity quota, reducing your available conversational allowances during the same window.
Terminal and CLI workflows: ChatGPT credentials vs. API keys
A widespread source of confusion among developers is how Codex operates within command-line interfaces (CLI) and terminal environments. There are two entirely distinct authentication paths, each governed by different billing models:
Path 1: ChatGPT Account Authentication (OAuth Web Login)
When utilizing OpenAI’s official developer CLI tools signed in via your browser account credentials:
- Your terminal commands consume your ChatGPT account allowance.
- On a Free plan, you can run occasional CLI-assisted commands using GPT-5.6 Terra until you hit your standard account rate limit.
- On a Plus ($20/mo) or Pro ($200/mo) plan, your terminal sessions leverage your expanded subscription quotas without triggering incremental per-token charges.
Path 2: API Key Authentication (OPENAI_API_KEY)
When passing a programmatic API key into developer tools, terminal scripts, or IDE extensions:
- Requests bypass your ChatGPT subscription entirely and query the OpenAI Developer Platform API.
- Usage is billed strictly on a pay-as-you-go per-token basis (covering input tokens, output tokens, and reasoning tokens).
- Having an active ChatGPT Plus or Pro subscription provides zero free API token credits. Even if you pay $20 or $200 a month for the consumer ChatGPT app, any CLI tool configured with an
OPENAI_API_KEYwill incur separate platform charges against your API billing credit card.
Rate limits, execution sandboxes, and context windows
Understanding Codex’s runtime boundaries helps engineering teams avoid unexpected pipeline failures:
- Sandbox isolation: Codex executes code within a sandboxed virtual container. Free accounts operate with restricted network access, preventing external package installations from untrusted third-party repositories during execution runs.
- Execution timeouts: Terminal commands and automated test runs in free sessions have strict execution timeouts (typically 30 to 60 seconds). Long build scripts, heavy database migrations, or resource-heavy compilation tasks will be terminated by the sandbox manager.
- Context window ceiling: While the underlying GPT-5.6 architecture supports large token contexts, free user sessions apply aggressive rolling summarization to preserve server resources. Large codebases with dozens of open files cannot be simultaneously held in active working memory on the free tier.
When to upgrade: ChatGPT Plus vs. Pro vs. Dedicated developer tools
Deciding whether to remain on free Codex or invest in premium tooling depends on your daily development intensity and technical requirements.
Tier | Monthly cost | Engine access | Codex allowance | Terminal / CLI access | Best suited for |
|---|---|---|---|---|---|
Free | $0 | GPT-5.6 Terra | Strict rate limits, lightweight execution | Basic browser sign-in | Learning programming, quick syntax checks |
ChatGPT Plus | $20 / mo | GPT-5.6 Sol & Terra | High allowances (~5x free volume) | Full account OAuth sign-in | Solo developers, daily web & script editing |
ChatGPT Pro | $200 / mo | GPT-5.6 Sol Max & Ultra | Unlimited standard Codex, maximum compute | High-priority dedicated terminal access | Professional engineers, tech leads, deep refactors |
OpenAI API | Usage-based | Full model catalog via API | Programmatic endpoints governed by org tiers | Direct script integration via API keys | CI/CD pipelines, automated PR bots, custom tooling |
Key reasons to upgrade from Free Codex
- Multi-file project context: Plus and Pro tiers preserve deep repository context, allowing Codex to trace function calls across modules rather than inspecting files in isolation.
- Flagship reasoning models: Upgrading unlocks GPT-5.6 Sol and Max reasoning effort, enabling the engine to analyze complex logic bugs, write robust end-to-end integration tests, and suggest comprehensive architectural refactorings.
- Continuous momentum: Professional software development cannot tolerate sudden 2-hour rate-limit pauses mid-afternoon. Paid subscriptions ensure predictable, dependable uptime.
When to consider alternative developer tools
While Codex inside ChatGPT is versatile, specialized coding platforms offer distinct advantages:
- If you require seamless inline tab-autocomplete inside your local editor, consider Cursor or GitHub Copilot.
- If your team relies on autonomous background agents that resolve GitHub issues and submit verified pull requests, specialized agent platforms like Windsurf provide tailored git integration.
For developers seeking an accessible, powerful code assistant without a financial commitment, ChatGPT’s free Codex tier delivers tremendous value. When development scale demands deep reasoning and uninterrupted execution, transitioning to Plus or Pro becomes a vital investment.
Evidence boundary
Official sources
Editorial guidance grounded in official product sources.
FAQ
Common questions
Can I use Codex for free?
Yes. Codex is included in ChatGPT Free, with plan-specific usage limits.
Does ChatGPT Go include Codex?
Yes. ChatGPT Go includes Codex and uses GPT-5.6 Terra for Work and Codex.
Is Codex CLI free?
The CLI is free to install. ChatGPT Free sign-in provides limited included usage; API-key sign-in is billed at standard API rates.
Are Codex credits the same as OpenAI API credits?
No. Eligible ChatGPT Work and Codex usage can share plan-side credits, but those credits do not pay API invoices. API-key usage remains separately billed.
When should a team pay for Codex instead of staying free?
Pay when task volume, workspace ownership, governance, admin controls, or predictable coding-agent capacity matters more than a free starting route.
Is Codex free if I bring an API key?
No. Bringing an API key moves cost to the API billing account behind that key; it does not make API-backed usage free.
Next steps
Take the next buying step
Use these next pages to confirm the plan, tool, or alternate route that fits once the spend boundary is clear.