Comparison

Codex vs Claude Code: which should your company standardise on?

Book a 30-minute call For engineers Public benchmarks Run a fair pilot

1AYM is an OpenAI Select Partner.

Written by Tayyeb Mahmud, founder and Principal Architect. Claude Certified Architect – Professional · LinkedIn profile of Tayyeb Mahmud

Vendor facts checked on 29 September 2026, and US dollar seat prices on 30 September 2026, against each vendor’s own pages.

The short answer: standardise the rules, not the tool

Both products do the same core job. They read a code repository, change files, run commands and hand a change back for review, and both run in a terminal, in an editor, in a desktop app and in the vendor's own cloud. The differences that matter to a business sit elsewhere: where the code runs, how an administrator enforces policy, which cloud provider carries the model traffic, and which vendor you already buy from.

Our view, from using both in our own engineering work: choose the one that fits your existing contract and your security model, write the instructions and rules once, and let a team use the other where a pilot on your repositories shows it does the better job.

Sources: [1] [9]

What each tool is

Both are described here from the vendors' own documentation, read on 29 September 2026. Neither vendor's description is a measurement, and the model line-ups behind both change often. OpenAI's DevDay announcements are the latest example, and our analysis of OpenAI DevDay 2026, the first post in our AI news and analysis, sets out what each one means for an enterprise.

Claude Code (Anthropic): Anthropic describes it as an agentic coding tool that reads your codebase, edits files, runs commands and works with your development tools. It runs in the terminal, in VS Code and JetBrains editors, in a desktop app and on the web, and it can run in CI through GitHub Actions or GitLab. Most surfaces need a Claude subscription or an Anthropic Console account; the terminal, VS Code and JetBrains clients can also run through third-party cloud providers.
Codex (OpenAI): OpenAI's coding agent, now documented as part of ChatGPT. It runs as the Codex CLI in a terminal, as an editor extension, in the ChatGPT desktop app, and as Codex cloud, which runs tasks in isolated cloud environments started from the web, GitHub, GitLab, Linear or Slack. It is included in ChatGPT plans from Free upwards, or it can run on an API key, which leaves out the cloud features.
What they share: Both read a project instruction file, both load skills written in the open Agent Skills format, and both connect to outside tools through MCP, run hooks and use subagents. That shared ground is why moving between them is cheaper than it used to be. Each is what we call an AI harness: the vendor builds and maintains it, and the instructions, skills and rules on top are the company's to write.

Sources: [1] [2] [3] [9] [10] [11] [13] [14] [18]

Which tool is for developers, and which for desk work

Both vendors now sell a second agent beside the coding tool, and the names make it easy to compare the wrong pair. Codex and Claude Code are the coding tools. ChatGPT Work and Claude Cowork are for desk work, such as a report or a slide deck. Put Codex against Claude Cowork, or Claude Code against ChatGPT Work, and you are comparing tools built for different people.

Codex and ChatGPT Work (OpenAI): OpenAI's help centre describes Work as an agent for longer, multi-step jobs that end in a finished document, spreadsheet, presentation or report. Codex is the one for software: writing and debugging code, running tests and commands, and working in a repository. In the ChatGPT desktop app, Codex has its own view and its own history. The two share usage, because Work runs on the same pricing, credits and usage limits as Codex. Workspace admins switch Work and local Codex access on and off separately.
Claude Code and Claude Cowork (Anthropic): Anthropic built Cowork as a separate place for bigger pieces of work, such as a weekly report or a slide deck. That is changing. On 16 September 2026 Anthropic announced that Cowork and chat are merging into one Claude app, rolling out to Pro and Max plans first, then Team and Free, with at least 30 days' notice to Enterprise admins before anything changes for them. Claude Code is the coding tool. It comes with every paid plan and shares the plan's usage limits, so work in the terminal and chats draw from one pool.

The buying decision is simpler than the naming. Give developers the coding tool and everyone else the desk-work agent, usually on the same plan, then size that plan for combined use. With either vendor, one person's coding and desk work come out of the same allowance, so a heavy week in the terminal leaves less for the weekly report.

Sources: [8] [18] [34] [35]

What public benchmarks say

You should not have to take a vendor partner's word on which tool writes better code, and this page is written by one, so it offers no head-to-head of its own. It points to the public benchmarks instead, each read on its maintainer's own page on 29 September 2026, and says who runs each one. Read the small print before the scores. Most of these rank models rather than products, and the question in front of you is about two products.

SWE-bench Verified (the SWE-bench team): 500 human-filtered tasks taken from real GitHub issues in Python projects, created with OpenAI in 2024. Its default view runs every model in the same minimal agent, mini-SWE-agent, and marks the runs the SWE-bench team performed or checked. No entry is Claude Code or Codex. The entries named after OpenAI's Codex models, such as GPT 5.2 Codex, are that model running in mini-SWE-agent, not in the Codex product. The newest entries were dated February 2026. On 23 February 2026 OpenAI said it had stopped reporting Verified scores, because an audit found material flaws in the tests or problem descriptions of 59.4% of the problems it reviewed, and signs that frontier models had seen some of the answers in training. This page quotes no Verified score.
SWE-Bench Pro (Scale AI): 1,865 tasks from 41 repositories, split into a public set of 731, a private set of 276 from startups' own code, and a held-out set whose scores are not published. Scale runs the models itself, in the SWE-Agent or mini-SWE-agent scaffold, and its leaderboard lists models, not Claude Code or Codex. Its GPT 5.2 Codex entry is that model in Scale's scaffold, not the Codex product. OpenAI recommended Pro in February 2026, then on 8 July 2026 estimated that about 30% of its tasks are broken and withdrew the recommendation. OpenAI's July post and Scale's live leaderboard give different top figures for the public set. When two primary sources disagree on a number, this page quotes neither.
Terminal-Bench 4.0 (Stanford, Harbor and Laude Institute): Tasks an agent carries out in a computer terminal. Version 4.0, released on 28 August 2026, recalibrated time and computing limits, removed 8 tasks and fixed 19. For this question it is the most useful of the four, because unlike the two above its entries include Claude Code and Codex by name, each paired with a model. On the leaderboard read on 29 September 2026, the top Codex entry (GPT-6 Astra at max reasoning, dated 3 September 2026) resolved 58.2% of tasks, with a 95% confidence interval of plus or minus 2.8 points, and the top Claude Code entry (Claude Fable 5.1 at max reasoning, dated 1 September 2026) resolved 57.9%, plus or minus 3.8 points. Those intervals overlap almost entirely. On this benchmark the two are level within its own margin of error, and anyone quoting a 0.3-point gap as a lead is reading noise.
METR's scaffold note (METR, a research non-profit): On 13 February 2026 METR compared Claude Opus 4.5 inside Claude Code, and GPT-5 inside Codex, with its own general-purpose agents on long autonomous tasks. Neither product did clearly better, and the differences were not statistically significant. METR points out that both products are built for interactive use with a person checking in, while its tasks run with nobody watching. The models it measured are also older than the ones in either product today.

Who ran what matters. Vendors publish scores for their own models at launch: OpenAI says it reports external and internal benchmarks with each model release, and when Anthropic put a Terminal-Bench figure on its Claude 4 model card in May 2025, the Terminal-Bench team said it would verify it before listing it. The Terminal-Bench team describes the 4.0 leaderboard runs as its own, and says its leaderboard experiments are supported by grants from OpenAI, Anthropic, Z.ai, SpaceX AI and Laude Institute.

Treat the top entries as a ceiling rather than a forecast for your seats. Both ran at max, the highest reasoning setting the leaderboard lists. Anthropic's documentation says neither Fable model is the default for any account type, and OpenAI says Plus and Business Standard seats include only limited GPT-6 Astra usage.

The bigger gap is what benchmarks leave out, which is most of what a business is buying. None of them measures admin controls, permission settings, audit trails, what seats and usage will cost your team, or how either tool behaves on your repositories with your reviewers. The numbers move as well. Terminal-Bench is now maintained as a continuous benchmark, so check the live leaderboard rather than this page. I would use a benchmark to narrow a shortlist and a pilot on your own code to decide.

Sources: [25] [26] [27] [28] [29] [30] [31] [32] [33] [34]

What switching costs

The licence is the small part. Both vendors list their entry business seat at US$20 per user a month billed annually, or US$25 billed monthly: a Claude Team standard seat and ChatGPT Business. The larger cost is people's time to move and re-check what your teams have built around the tool: instruction files, skills, permission settings, hooks and connections to internal systems.

That cost has fallen, because each tool can now import the other's setup. OpenAI's import flow, in the Codex CLI and the ChatGPT desktop app, reads a Claude Code setup and brings across instruction files (as AGENTS.md), settings.json (as config.toml), skills, plugins, MCP server settings, hooks, slash commands (as skills), subagents, project memories and recent chats; the CLI takes up to 50 chats from the last 30 days. It leaves the Claude Code setup unchanged.

In the other direction, Claude Code's /import command, from version 2.1.213, reads a Codex setup on the same machine. It appends a one-time copy of instruction files such as AGENTS.md to the matching CLAUDE.md and brings across MCP servers, commands, subagents and skills, and a dry run shows what it would change before it writes anything. When /init finds Codex configuration, it offers to run the import. The command is not available when Claude Code runs through Amazon Bedrock, Google Cloud's Agent Platform, Microsoft Foundry or Claude Platform on AWS. Without it, Claude Code still reads a repository's AGENTS.md when the repository has no CLAUDE.md.

What does not move by itself is judgement. OpenAI's own import guidance says to review imported permissions and tool restrictions, MCP servers that use custom authentication, hooks whose behaviour may differ, and prompts that depend on arguments or file paths. The two permission models are also built differently, so an administrator's policy has to be rewritten rather than converted, whichever way the move goes.

Separate migration and review effort from the ongoing bill: see how Codex plans, usage and credits work before estimating the cost of a switch.

Sources: [2] [8] [12] [18] [20] [24]

Do we need native import, delivery support or a 1AMS pilot?

Start with the work your team needs to complete. Native import may cover the transfer. Choose 1AYM when the harder part is integrating the tools, proving the workflow and rolling it out. A 1AMS pilot lets you assess the product against the reusable assets you want to manage across tools.

When to choose native import, 1AYM delivery support or a guided 1AMS pilot
Dimension	Choose it when	What to establish
Native import	Your team can transfer supported setup and validate the result itself.	Review imported permissions, connections, hooks and task behaviour. OpenAI also documents automatic updates in the desktop app. Check the supported import flow in source [12].
1AYM delivery support	You need architecture, integration, testing or rollout capacity around an existing engineering team.	Scope the inventory, acceptance tests, repairs, ownership and rollout decision. We build the skills, guardrails and review practices your engineers take forward. Explore AI harness and platform engineering for the delivery scope.
1AMS guided pilot	You want to evaluate managing reusable skills, agents and plugins across vendor tools.	See the current demonstration, agree the supported tools and assets, then test the fit on a defined workflow. Explore the 1AMS pilot and use those results to decide what comes next.

Native import and automatic updates checked 5 October 2026. Delivery support and 1AMS are scoped to the engagement or pilot agreed with your team.

A successful import is one result to check. The buying decision also depends on which tasks work, which controls hold, how much repair and review remain, and who owns the setup afterwards.

Once you have chosen a route, plan the team rollout, ownership and checks before expanding beyond the first team.

Download the Claude Code to Codex migration evidence pack for actual local converter output and reproducible fixture checks. Native import, runtime controls, task quality and time savings remain unmeasured.

Sources: [12]

Usage on top of seats

Seats are the easy half. Put 20 developers on each vendor's entry business plan, billed annually, and both come to US$400 a month, or US$4,800 a year, before anyone uses anything (our arithmetic). On cost, the Codex vs Claude question is a usage question, and the two vendors meter usage differently.

Claude Code and Codex for 20 developers: the business plan each comes in, seat prices, the monthly seat bill for 20, what a seat's usage covers and what happens past it
Dimension	Claude Code (Anthropic)	Codex (OpenAI)
Business plan it comes in	Claude Team, for 2 to 150 seats. Claude Code is in every paid plan	ChatGPT Business, from 2 seats. Codex is in every ChatGPT plan
Standard seat, a month	US$20 billed annually, US$25 monthly. UK: £18 or £22.50	US$20 billed annually, US$25 monthly. UK: £15 or £18
20 standard seats, a month	US$400 billed annually, US$500 monthly. UK: £360 or £450	US$400 billed annually, US$500 monthly. UK: £300 or £360
What a seat's usage covers	One allowance per seat that resets every five hours and every week, shared with Claude chat and Cowork	One allowance per seat, shared with ChatGPT Work. OpenAI estimates 15 to 150 local GPT-6 Sol messages per five hours on a standard seat, and weekly limits may apply
Premium seat for heavy users, a month	US$100 billed annually, US$125 monthly, for five times a standard seat's usage. UK: £90 or £113	OpenAI's help centre names Business Premium seats, which can spend their whole allowance on GPT-6 Astra (checked 29 September 2026). For the current price, see OpenAI's Business pricing page
Past the allowance	Extra usage at standard API rates, once an admin turns it on, with spend limits per user and for the organisation	Workspace credits, priced in credits per million tokens. The price of a credit is on your rate card
Enterprise route	Claude Enterprise: US$20 a seat a month, billed annually, plus all usage at API rates. At least 20 seats	ChatGPT Enterprise: quote-only. On flexible pricing, usage scales with credits
Published usage reference point	About US$150 to US$250 per developer a month across enterprise deployments: US$3,000 to US$5,000 for 20 developers (our arithmetic)	No per-developer figure. OpenAI's one per-task figure is for GPT-5.6 Sol: 5 to 30 credits a task

US dollar prices read on the vendors' pages on 30 September 2026. Pound prices as shown to a UK buyer on 29 September 2026: Anthropic's include 20% VAT, OpenAI's page does not say. Anthropic's US prices exclude tax. Totals are our arithmetic.

Usage is harder to pin down. Anthropic publishes one reference point for Claude Code: across enterprise deployments, an average of about US$13 per developer per active day and US$150 to US$250 per developer per month, with 90% of users below US$30 per active day. OpenAI's Codex pricing page describes usage as plan limits, then credits priced per million tokens, with a typical credit range per task, and gives no per-developer figure. Neither is a forecast for your team. A two-week pilot on your own repositories is.

The pilot also shows who needs a bigger seat. On Claude Team a premium seat is priced at five times a standard one, so budget from how many of your 20 developers run out of allowance in a normal week, not from the average.

Prices and plan limits change. The US dollar seat prices were re-read on 30 September 2026; check both pricing pages on the day you buy.

For the plan-by-plan prices at business and enterprise level, see our ChatGPT Enterprise vs Claude Enterprise guide, which compares every tier on the vendors' own pricing pages.

Sources: [7] [8] [18] [34] [39] [40] [41]

Governance: where each tool keeps its controls

For a security reviewer the useful question is not which tool is safer in general. It is where the controls live, who can change them and what happens by default.

Default behaviour: Claude Code's Manual mode asks before most edits, commands and network access. From version 2.1.283 its starting mode in the terminal and VS Code is auto mode, in which a second model reviews actions instead of a person. A reviewing model is still a model, and our pattern note sets out why a model should not be the only gate on a consequential action. Codex runs with network access off by default, and for a version-controlled folder OpenAI recommends its Auto preset, which edits and runs commands inside the project and asks before going outside it or onto the network.
Organisation policy: Claude Code applies a managed settings file above every user and project setting; it can deny specific file reads and commands, turn off the mode that skips all checks, and deploy approved MCP servers. Codex enforces administrator requirements that users cannot override, covering the approval policy, the sandbox, web search, managed hooks and which MCP servers people can switch on.
Where the code runs: Both work on the engineer's own machine, and both offer sessions that run in the vendor's cloud: Claude Code on the web, and Codex cloud, whose environments are isolated containers with the agent working offline unless internet access is switched on.
Which cloud carries model traffic: Claude Code can route through Amazon Bedrock, Google Cloud's Agent Platform or Microsoft Foundry. Codex's local clients can use OpenAI models through Amazon Bedrock with AWS authentication, with OpenAI's hosted API out of the request path, and its configuration documents an Azure OpenAI provider. If your data policy names a cloud provider, this can settle the question on its own.
Training on your data: Both vendors' business plans state no model training on your content by default: Claude's Team plan and ChatGPT Business.

Sources: [1] [4] [5] [6] [8] [11] [15] [16] [17] [18] [19]

A decision guide in five questions

Answer these in order. The first firm answer usually decides it.

Do you already buy ChatGPT Enterprise or Claude Enterprise?: Start with the tool inside the contract you have. Sign-on, audit, retention and procurement are already settled there, and both tools sit inside those plans.
Does your data policy name a cloud provider?: If model traffic must stay with Google Cloud, Claude Code is the one with a documented route there: as of 29 September 2026, OpenAI's Codex configuration docs list no Google Cloud provider. Both document a route through Amazon Bedrock, and on Bedrock both lose their cloud features. Codex runs locally only, without Codex cloud, its GitHub, Slack and Linear cloud integrations, web search or admin features such as SAML SSO, SCIM and the analytics API. Claude Code runs without cloud sessions, Claude Code in Slack, Remote Control, web search, admin features such as the analytics dashboard and server-managed settings, or the /import command. Both can use Microsoft's cloud: Claude Code through Microsoft Foundry, Codex through an Azure OpenAI provider.
Do you want agents working unattended in the vendor's cloud?: Both offer it. Read how each isolates the environment, handles secrets and controls internet access before you allow it on a real repository. If the code is on GitHub, GitHub's own Copilot cloud agent is a third route, and our comparison of Codex and GitHub Copilot sets out how it differs.
Is the engineering team already split between the two?: Then do not force a switch. Keep the instructions in one AGENTS.md, keep skills in the shared format, and hold one written permission policy per tool. The rollout guide for Codex and Claude Code, in the list of guides on this topic at the end of this page, covers that operating layer in full.
Which one does better work on your code?: Brochures and leaderboards cannot answer that for your code. A pilot can. Run both for two weeks on one repository that ships often, and count merged changes, reviewer time and rework. The next section sets out how to keep that comparison fair.

If you keep both, the skills and rules are what you end up maintaining twice, so write them once in the shared formats and keep them in the repository.

Sources: [6] [8] [11] [15] [17] [18] [19] [20] [24]

How to run a fair pilot of your own

Vendor documentation says what each tool can do. It cannot say how each behaves on your code, with your reviewers. A two-week pilot can, as long as both tools get the same conditions.

The same starting point: Pick one repository that ships often and has tests you trust. Keep one AGENTS.md as the instructions for both tools and no project CLAUDE.md, so both start from the same file.
The same conditions: Run each tool on your business plan with its default model, give both the same written task, and fix a time limit and a cap on follow-up messages before anyone starts.
A clean configuration: Start each run without anyone's personal instructions, skills, MCP servers or plugins, and with memory features off, so one engineer's setup does not decide the outcome.
A blind review: Have a reviewer judge each change as a diff, with commit messages and attribution lines removed, before they learn which tool wrote it. Claude Code adds a Co-Authored-By line to its commits by default.
What to count: Merged changes, reviewer minutes to a merge decision, the rework the reviewer asked for, and usage as each tool reports it. Write down the tool versions and the dates.

Check the controls as well as the code. Give each tool a task that leads towards something your policy forbids, such as reading a secrets file or pushing to a protected branch, and record whether it asked first, was blocked or went ahead. Two weeks on one repository will not settle which tool is better in general. It will show which one suits your team on your code, which is the decision you are making.

Sources: [2] [5] [13] [16] [21] [22] [23]

For engineers: configuration, controls and a clean pilot setup

Instruction files

Claude Code reads CLAUDE.md, .claude/CLAUDE.md and CLAUDE.local.md up the directory tree, plus the user's ~/.claude/CLAUDE.md and any managed-policy CLAUDE.md. It reads AGENTS.md only when none of those three project files exists in the working directory or above it, unless the Project instructions setting says otherwise. Reading AGENTS.md directly needs Claude Code 2.1.277 or later.

Codex reads AGENTS.override.md or AGENTS.md from its home directory, then from the project root down to the working directory. Files are joined root first, so the nearest file wins, up to 32 KiB by default (project_doc_max_bytes).

For a repository both tools use, keep AGENTS.md as the source. Either leave out a project CLAUDE.md, or have CLAUDE.md import AGENTS.md.

Sources: [2] [13]

What each import maps

Codex, from Claude Code: instruction files to AGENTS.md. settings.json to config.toml. Slash commands to skills. MCP configuration, hooks and subagents to their Codex equivalents. Project memories to memories. The CLI command is /import; it is not available during a running task, in a remote session or while connected to a local app-server daemon.

Claude Code, from Codex: /import codex (the same command also takes gemini or cursor). Instruction files such as AGENTS.md are appended once to the matching CLAUDE.md; MCP servers, commands, subagents and skills come across too. --dry-run previews without writing anything and --yes skips the interactive picker; with -p it lists what it found and prints the command that confirms the import. It needs v2.1.213 or later, and it is not available on Amazon Bedrock, Google Cloud's Agent Platform, Microsoft Foundry, Claude Platform on AWS, through a Claude apps gateway, or with feature-flag fetching turned off. /init offers it when it finds Codex configuration.

Sources: [2] [12] [20]

Skills in both

Both follow the open Agent Skills format: a directory with a SKILL.md file and optional scripts and references. Codex requires a name and a description in the frontmatter. Claude Code adds frontmatter fields of its own, so for a skill that must load in both, keep the frontmatter to the six fields in the open specification.

Sources: [3] [14]

Permission models

Claude Code has permission modes: default (shown as Manual), acceptEdits, plan, auto, dontAsk and bypassPermissions. Allow, ask and deny rules sit on top, and deny rules apply in every mode.

Codex combines a sandbox mode (read-only, workspace-write or danger-full-access) with an approval policy such as on-request. The untrusted approval policy is retired. From Codex 0.138.0, permission profiles are the preferred way to set this.

Sources: [4] [15] [16]

Enforcement

Claude Code: managed-settings.json, an MDM profile or server-managed settings, applied above user, project, local and command-line settings. The documented example denies Read(./secrets/**) and disables bypass mode.

Codex: requirements.toml (for example /etc/codex/requirements.toml) constrains approval policy, sandbox mode, permission profiles, web search, managed hooks, allowed MCP servers and plugin sources. Managed config.toml values are defaults that users can override; requirements are not.

Sources: [5] [16]

Setting up a clean pilot

Give each run a new, empty configuration directory: CLAUDE_CONFIG_DIR for Claude Code, CODEX_HOME for Codex. Claude Code keeps settings, session history and plugins under its directory, and Codex reads a user-level AGENTS.md from its home, so an empty one keeps personal setup out of the run.

Claude Code's auto memory is on by default and shared by every worktree of the same git repository; CLAUDE_CODE_DISABLE_AUTO_MEMORY=1 turns it off. Codex's local memories are off by default. Do not run either import command during the pilot, or the second tool starts from the first tool's setup.

For the blind read, export the diff from the starting commit to the final working tree, untracked files included and tool directories such as .claude/ and .codex/ excluded. Strip commit messages and trailers, including Claude Code's default Co-Authored-By line, before the reviewer sees it.

For the policy check, put the same restrictions in each tool's enforcement file, managed-settings.json for Claude Code and requirements.toml for Codex, and record the tool version, model, plan and operating system for every run. Claude Code reports usage with /usage.

Sources: [2] [5] [7] [13] [16] [21] [22] [23]

Where 1AYM fits

Setting up and rolling out these tools for companies is core 1AYM work: Codex and Claude Code for engineers, ChatGPT Work and Claude Cowork for everyone else, and the instructions, skills, permissions and CI gates around them. The documented case is engagement file D-01, which shows what re-architecting a skills estate on a production platform changed. Whichever tool you pick, that operating layer is our AI harness and platform engineering work. If you keep both, 1AMS, our product for writing skills once and deploying them to each vendor's own tools, is offered as a guided pilot.

It does not have to start big. 1AYM takes small fixed-scope statements of work as well as larger builds, and fixed-scope work can start within a day of the scope being signed.

Sources

Frequently asked questions

Is Codex or Claude Code better for a business?

Neither in general. For a business the deciding factors are the contract you already hold, where your policy lets code and model traffic run, and how each tool behaves on your own repositories. Both list their entry business seat at the same price. Run both on one repository for two weeks and compare merged changes, reviewer time and rework.

Can we move from Claude Code to Codex without starting again?

Largely, yes. OpenAI's import flow in the Codex CLI and the ChatGPT desktop app brings across instruction files, settings, skills, plugins, MCP settings, hooks, slash commands, subagents, project memories and recent chats, and leaves the Claude Code setup unchanged. OpenAI says to review permissions, custom-authenticated MCP servers, hooks and argument-dependent prompts after importing.

Can we move from Codex to Claude Code?

Largely, yes. Claude Code's /import codex command, from version 2.1.213, appends a one-time copy of instruction files such as AGENTS.md to CLAUDE.md and brings across MCP servers, commands, subagents and skills; a dry run shows what it would change first. It is not available when Claude Code runs through Amazon Bedrock, Google Cloud's Agent Platform, Microsoft Foundry or Claude Platform on AWS. Without it, Claude Code still reads a repository's AGENTS.md when there is no CLAUDE.md. Permission and administrator policy have to be rewritten, because the two tools model permissions differently.

Claude vs Codex: which is better for coding?

No public source settles it, and I would be wary of anyone who says one does. On Terminal-Bench 4.0, read on 29 September 2026, the top Codex and Claude Code entries were within each other's margin of error. SWE-bench Verified and SWE-Bench Pro rank models rather than these two tools, and OpenAI has since questioned both. METR found that, for the models it measured, running them inside Codex or Claude Code made little difference. Which one does better on your code is a question only a two-week pilot on your own repository can answer.

Codex vs Claude: which costs more for 20 developers?

Neither, on seats. Claude Team and ChatGPT Business both list a standard seat at US$20 a month billed annually, so 20 seats cost US$400 a month with either vendor (our arithmetic, prices read on 30 September 2026). The difference is usage: how many developers outgrow a standard seat's allowance (on Claude Team, a premium seat costs five times a standard one), and how much extra usage or credit the rest buy past their allowance. A two-week pilot on your own repositories shows both before you commit.

Where does GitHub Copilot fit?

GitHub Copilot is GitHub's own set of coding tools: a cloud agent that works in a GitHub Actions environment and prepares pull requests, agent mode in the editor, and a CLI. It is not tied to one model vendor. GitHub's supported-models list, read on 29 September 2026, includes OpenAI models such as GPT-6 Sol and Anthropic models such as Claude Opus 5.5, and GitHub lists Anthropic Claude and OpenAI Codex as third-party coding agents on GitHub, in public preview on paid Copilot plans. None of the public leaderboards above lists Copilot as an agent. If you already pay for Copilot, put it in the same pilot as a third option rather than judging it on its own.

Why is the comparison written by an OpenAI Select Partner?

Because that is who we are, and the page says so at the top for that reason. Every fact here comes from a vendor's own documentation or a benchmark maintainer's own page, dated and listed in the sources, so anyone can check it. The page offers no head-to-head of its own. The pilot section shows how to compare both tools on your repositories without relying on our word.

Sources for “Claude vs Codex: which is better for coding?”: [25] [26] [27] [28] [29] [32]

Sources for “Codex vs Claude: which costs more for 20 developers?”: [8] [18]

Sources for “Where does GitHub Copilot fit?”: [36] [37] [38]

In this topic

Which coding agent should our engineers use, what will it cost and how do we roll it out?

Rolling out coding agentsGuide
How do you roll out Codex and Claude Code across an engineering organisation?
Codex pricingPricing guide
How much does Codex cost, and what will it cost our team?
Claude Code vs GitHub CopilotComparison
Should an engineering team on GitHub use Claude Code or GitHub Copilot?
Codex vs GitHub CopilotComparison guide
Should a team whose code is on GitHub use Codex, GitHub Copilot or both?
How to write AGENTS.mdGuide
How do I write an AGENTS.md or CLAUDE.md file for our codebase?

Further

AI platform enablement · The engagement file behind our delivery experience and skills-estate work.
What is an AI harness? · What the vendor runtime provides and what the company owns around it.
Agent proposes, verifier gates · Why consequential actions need an independent check before they ship.

Choosing, or running both?

Half an hour is enough to tell you which way we would go on your repositories, what the operating layer would take, and roughly what it would cost.

Book a 30-minute call enquiries@1aym.com See the engagement files

Last reviewed 5 October 2026 · 1AYM