./compare/ claude-code-vs-codexfacts checked 17.08.2026

Claude Code or Codex: which to choose

Short answer: only Codex can be tried for free, and it is included even in the ChatGPT Free plan. Claude Code is chosen for Claude models and up to 1M context. Next, we break down where this difference actually changes the work.

At a glance

Claude CodeCodex

Entry price

from $17/mo

Pro with annual billing

✓from $0

Included in all ChatGPT plans

Free plan

No

The Free plan does not provide access

✓Yes

In the ChatGPT Free plan

Context window

✓200K–1M

1M with Fable 5, Sonnet 5, and Opus 4.6+

—

Not documented in the fact pack

Open source

No

Proprietary CLI

✓Yes

CLI Apache-2.0, ~90k stars

Surfaces

✓14

CLI · IDE · desktop · web/iOS · CI · Slack

11

app · CLI · IDE · cloud · GitHub · mobile

Access from Russia

No

Russia is not on the supported list

No

Not in supported countries

These are different form factors, and that is the key point

Claude Code was born as a terminal agent and still lives primarily in the terminal: it grows into an existing repository, reads CLAUDE.md, retains project memory, runs tests, and prepares pull requests. Around its core, 14 surfaces have grown: CLI, extensions for VS Code and JetBrains, a desktop app, web and iOS, CI integrations, Slack, SDK. But its nature is unchanged: it is a tool for serious work in existing code, not autocomplete.

Claude Code · terminal

Codex is built the other way around: it is one product with several windows into it. It lives in the ChatGPT app, in a desktop app with parallel threads and worktrees, in CLI, an IDE extension, the cloud, GitHub review, and even on a phone. Notable features include computer use on macOS and Windows, cloud tasks with best-of-N, scheduled automations and Sites, and website deployment on OpenAI hosting.

Codex · app

The practical takeaway: if your work is one large repository and deep changes within it, Claude Code is a better fit. If you want a pipeline of parallel tasks of different kinds from any device, Codex is a better fit.

Only Codex lets you start for free

Codex is included in every ChatGPT plan, including Free: no card required, limits depend on the plan. Paid tiers: Plus $20, Pro from $100 per month. Claude Code has no free entry: the Free plan does not provide access, entry starts at $17 per month for Pro with annual billing, and heavy workloads live on Max for $100-200.

The limits work similarly: both have windows of about 5 hours, while Claude Code also has weekly bars with a shared pool across claude.ai, Claude Code, and Desktop. An important historical detail: on 27.05.2026, Claude Code limits were depleted faster because of a cache regression; Anthropic rolled back the change and reset limits for everyone. This happens with both vendors: limits are a living system, not a constant from the price list.

Numbers duel: look at the AA Index, not promises

Comparing agents in words is pointless; that is what the AA Coding Agent Index is for: price and time on real tasks, each agent-model pair gets its own score. Latest snapshot from 23.09.2026:

Read it this way: the score belongs to the pair, not the agent in general. Changing the model inside an agent moves the result more than changing the agent itself, and the leader flips with major model releases. The full ranking and methodology are open: AI agent ranking.

Models and context: Claude Code's trump card

Claude Code runs on Claude models: Fable 5, Opus 4.8, Sonnet 5. The context is 200K tokens, and up to 1M for Fable 5, Sonnet 5, and Opus 4.6+: a window that fits a substantial part of a medium-sized repository without workarounds. Codex uses GPT-5.6 as its main model, alongside GPT-5.5, GPT-5.4, and GPT-5.4-mini; we found no comparable context claim in its materials, and honestly record this as “not specified,” rather than a loss.

Codex has open source code instead: its CLI is on GitHub under Apache-2.0, written in Rust, with around 90 thousand stars. For teams that need to read and patch the tool, this is a real advantage: the Claude Code CLI is proprietary.

What about access from Russia

There is parity here, and it is unpleasant: Russia is not on either supported-country list. We cover working payment and access routes in a separate section: pay for AI from Russia. There is no direct Russian equivalent for this pair; the closest local option is SourceCraft Code Assistant from Yandex: agent mode from task to pull request, a Free plan, and Pro for 250₽ per month, payable in rubles.

What we build this site with

Agents vibecode vibecoding.tech; a human writes the rules and accepts the work. As of September 8, 2026, nine-tenths of the site was made in Claude Code and one-tenth in Codex: of 276 merged changes, 250 were by the former and 26 by the latter.

The roles split according to each one's strengths. Claude Code builds pages, data, and checks; Codex reviews news before publication and generates images. What is happening right now is shown by live machine feed.

Which to choose

What changed in recent months