Best AI tools for developers: how to shortlist

admin
Best AI tools for developers: how to shortlist cover graphic

AI coding tools used to compete on one thing: how good the autocomplete was. Today they differ on where they run (editor, terminal or cloud), how they charge (flat fee, credits or weekly usage pools) and how much work they will do without you watching. That makes “which is best” a poor question. “Which fits how I work and how I want to be billed” is a better one.

This guide gives you a shortlisting method, then sets out documented facts about four widely used options: GitHub Copilot, Cursor, Claude Code and OpenAI’s Codex. We compare their pricing models, usage limits and data terms. We do not rank code quality, because we have not run these tools on code for this article.

How we researched this: we did not run hands-on tests for this article. It is based on the vendor’s own documentation and public sources, checked on 11 November 2026. Features and prices change, so confirm details on the vendor’s site before you buy.

Key takeaways

  • Pricing models have moved beyond flat fees: GitHub says its plans use AI credits (1 credit equals $0.01), Cursor describes weekly-refreshing usage pools, and Anthropic and OpenAI describe session and weekly usage limits.
  • GitHub’s plans page lists Copilot Free at $0 with 2,000 completions a month, Pro at $10, Pro+ at $39 and Max at $100 per user per month.
  • Cursor’s pricing page lists Plus at $8, Premium at $30, Super at $100 and Ultra at $200 per month, with third-party models billed at API list price.
  • Claude Code is included with Claude Pro, Max and Ultra plans according to Anthropic’s support page, and shares its limits with Claude chat.
  • Data terms vary: Cursor says it will not train on your data when Privacy Mode is enabled, and GitHub’s docs carry a footnote that some Anthropic models retain data by default for safety classifiers.

Why shortlist by workflow

Developers use AI in at least three modes: completions while typing, chat or edit commands inside an editor, and agents that take a task and work through it across files. A tool that excels in one mode may be average in another, and the billing for each mode can differ. GitHub’s page, for example, says code completions do not use credits while chat, agent mode, code review and CLI do.

Start from your constraints

Decide four things before you open a pricing page: your editor or terminal habits, whether code can leave your machine under your employer’s policy, how much unattended agent work you want, and your monthly budget ceiling. For the agent question, our guide on choosing between a chatbot and an AI agent may help.

Quick overview table

All details are from the vendor pages we read on 11 November 2026.

Tool Individual plans listed How usage is limited
GitHub Copilot Free $0; Pro $10; Pro+ $39; Max $100 per user per month; free Student plan for verified students Monthly AI credits (1 credit equals $0.01); completions unlimited on paid plans
Cursor Free $0; Plus $8; Premium $30; Super $100; Ultra $200 per month Usage pool per tier, refreshed weekly; top-offs available
Claude Code Included with Claude Pro ($17 per month annual, $20 monthly), Max ($100) and Ultra ($200) Five-hour session limit and weekly limit shared with Claude chat
OpenAI Codex Free $0; Go $8; Plus $20; Pro from $100; Business $20 per user annual or $25 monthly Estimated messages per five-hour period on Plus; credits for extra usage

Criteria that matter

  1. Where it runs. Editor extension, standalone editor, terminal, cloud agent.
  2. Billing model. Flat, credits, weekly pool, or API-rate billing.
  3. Model choice. Which models you can pick and whether they cost extra.
  4. Data handling. Training, retention, privacy modes.
  5. Team controls. SSO, audit logs, admin limits.
  6. Fit with your repo and review habits. Needs your own trial.

GitHub Copilot

Plans

GitHub’s plans page lists Free at $0 with 2,000 completions per month, a CLI, agent mode and community support, with limited chat and agent usage. Pro is $10 per user per month with unlimited completions and next edit suggestions, cloud agent, code review and model selection. Pro+ is $39 and Max is $100, with larger included usage. GitHub’s docs give monthly AI credits for each: 1,500 for Pro, 7,000 for Pro+ and 20,000 for Max, and describe a Student plan that is free for verified students with Auto model selection only.

Usage rules

The plans page says one AI credit equals $0.01. When your allowance runs out you can wait for the monthly reset, set a dollar budget for extra usage or switch to a cheaper model, and alerts fire at 75%, 90% and 100% of a budget. Purchasing extra credits is available on Pro, Pro+ and Max but not on Free. The page itself contains a few inconsistencies, such as whether third-party agent delegation is included on Pro, so check the live page.

Cursor

Cursor’s pricing page lists Free ($0, limited agent and chat requests), Plus ($8, with only Grok models available), Premium ($30, 4x the Plus usage, cloud agents and third-party models at API list price), Super ($100, 15x) and Ultra ($200, 50x). Teams start at $30 per user per month, and Enterprise is custom. Each paid tier has a usage pool shared across Cursor and Grok, and the page says the allowance refreshes weekly so one heavy day does not exhaust the month. Grok API usage is billed separately, and third-party models in Cursor are billed by usage at API list price. Prices exclude taxes, and subscriptions are sold only through Cursor’s website.

Claude Code

Anthropic’s support page says you can sign in to Claude Code with the same credentials you use for Claude, and that Pro, Max and Ultra plans include it, including inside VS Code, Cursor and JetBrains IDEs. Pro and Max have a five-hour session limit and a weekly limit; Ultra has a weekly limit, with no five-hour limit for verified accounts. Claude and Claude Code draw from the same pool. If you hit a limit you can upgrade, enable usage credits, switch to Console API credits or wait. The page also warns that if an ANTHROPIC_API_KEY variable is set, Claude Code uses it and you are billed at API rates instead of using your subscription. The pricing page lists Pro at $17 per month billed annually or $20 monthly, Max at $100 and Ultra at $200.

OpenAI Codex

OpenAI’s Codex pricing page, which shares pricing with ChatGPT Work, lists Free at $0 for quick coding tasks, Go at $8, Plus at $20 (adding Codex on the web, CLI, IDE extension and iOS), Pro from $100 with tiers at $100, $200 or $500, and Business at $20 per user per month billed annually or $25 monthly. API key use is billed at API rates and excludes some cloud features. For Plus, the page gives estimated local messages per five-hour period by model, and says these are estimates, not fixed limits. It also says Pro plans currently have no five-hour limit. We read the first part of a long page, so details further down may differ.

Other tools we could not read

We were unable to read official pages for several other tools, such as JetBrains AI and Google’s Gemini Code Assist, so we do not describe them. If they are on your list, apply the same criteria.

Pricing models compared

Checked on 11 November 2026. Confirm on each vendor page.

Model Examples Risk to watch
Flat fee plus included credits GitHub Copilot Pro, Pro+, Max Chat and agent work draws down credits; completions do not
Usage pool, weekly refresh Cursor paid tiers Third-party models billed at API list price
Shared subscription limits Claude Code with Claude plans; Codex with ChatGPT plans Heavy coding can use up limits shared with chat
API-rate billing Claude Code with API key; Codex with API key Costs scale with usage and have no plan cap

A sensible approach is to estimate a typical week, for instance the number of agent runs, and check the vendor’s own cost guidance. How to read an AI tool’s pricing page covers credits and overage terms.

Privacy and data

Tool What we could confirm
Cursor Privacy Mode can be enabled in settings or by a team admin, and Cursor states that when enabled it will not train on your data. The page does not state retention periods.
GitHub Copilot The plans documentation makes no explicit statement about training on individual users’ data. A footnote says that with some Anthropic models, Anthropic retains prompts and outputs by default to operate safety classifiers, with time-bound zero data retention exemptions for approved enterprises.
Claude Code Anthropic says Team plans include no model training on content by default; we did not confirm individual-plan terms.
OpenAI Codex OpenAI says consumer content may be used for training by default, with business offerings excluded; Enterprise lists data retention and residency controls.

Before pointing any tool at a private repository, check your employer’s policy and the plan’s data terms. Our guide on protecting company data when using AI tools sets out a checklist. Never paste secrets, keys or customer data into prompts.

Who each suits

  • GitHub Copilot: best for developers who live in an editor and want completions plus optional agent features, if credit-based billing is acceptable.
  • Cursor: best for developers who want an AI-first editor and clear weekly usage pools, if they will manage model costs.
  • Claude Code: best for developers comfortable in the terminal who already use Claude, if shared session and weekly limits fit their workload.
  • OpenAI Codex: best for developers already on ChatGPT who want coding tasks across web, CLI and IDE, if estimated usage fits.

Many developers use two tools together, for example an editor assistant and a terminal agent. That is a cost decision, not just a feature one.

How to run your own test

  1. Choose three real tasks. A bug fix, a small feature and a refactor, in a non-sensitive repository.
  2. Give each tool the same brief. See how to write a brief that AI can follow.
  3. Run tests and read every diff. Treat output as a pull request from a stranger. Use how to check AI output for mistakes.
  4. Record cost and limits. Note credits or usage consumed and any limit you hit.
  5. Compare after a week. Judge on time saved, review effort and surprises.

The general approach is in how to test an AI tool before you pay for it.

Decision checklist

  • I know whether I want completions, chat, agents or all three.
  • I have priced a typical week of usage, not just the plan fee.
  • I have checked whether my employer allows the tool and which data it sends.
  • I have turned on privacy settings where available.
  • I have tested on non-sensitive code and reviewed every diff.
  • I know how to cap spending or leave the plan.

Frequently asked questions

Which AI coding tool is best?

We cannot say, because we did not test code quality. The best choice depends on workflow, billing tolerance and data rules.

Is GitHub Copilot free?

GitHub lists a Free plan at $0 with 2,000 completions per month and limited chat and agent usage, and a Student plan that is free for verified students.

Does Claude Code need a separate subscription?

Anthropic says Pro, Max and Ultra plans include it, with limits shared with Claude chat. Using an API key instead bills at API rates.

Will these tools train on my code?

It depends on tool, plan and settings. Cursor states it will not train when Privacy Mode is enabled; for others, read the current terms for your plan.

Why does my usage run out quickly?

Agent work and premium models consume more of a credit or usage allowance than completions. Check the vendor’s guidance and set budgets.

Conclusion

A good developer shortlist starts with workflow and billing tolerance, then checks data terms, then tests on real tasks with every diff reviewed. On the documentation we read, the main differences are how each tool meters usage and what it says about data, so those deserve most of your attention. Pair this guide with how to choose an AI tool and how to compare AI tools side by side fairly, and verify details on each vendor’s site.

Sources

Keep reading