Cursor vs ChatGPT vs Claude: which AI for coding?

Don't pick a brand, pick the job. All three help you write and ship code with AI, but they do different work: Cursor is a repo-aware AI code editor with agents, ChatGPT is a general assistant that covers light coding, and Claude is strong on long context and coding (Claude Code on paid plans). This page is about repo / coding-agent jobs — for general chat assistants see ChatGPT vs Claude vs Gemini; for builders vs editor see v0 vs Lovable vs Cursor or Bolt vs Lovable vs v0; for IDE plugin vs AI-native editor vs Claude see GitHub Copilot vs Cursor vs Claude. Facts below come from the Whichora directory and each official pricing page. Prices change, so re-check before you pay. Not financial or legal advice. Last checked 2026-10-16.

← Back to Whichora · All guides · AI for coding · v0 vs Lovable vs Cursor · ChatGPT vs Claude vs Gemini · Bolt vs Lovable vs v0 · GitHub Copilot vs Cursor vs Claude

Side by side

CursorChatGPTClaude
Best forCoding inside a repo with AI agents (local and cloud)General assistant: writing, ideas, images, voice, light codingLong documents, careful reasoning, coding (Claude Code on paid plans)
Free tierYes (limited agent and chat requests)Yes (limits on uploads, images, voice, research, Codex)Yes (web/desktop/mobile chat, web search, files, Artifacts)
Starting paid pricePlus USD 8/mo (Grok models only) · Premium USD 30/mo · Super USD 100/mo · Ultra USD 200/mo; Teams from USD 30/user/mo. Third-party models billed separately at API list price. Plans are sold bundled with X / Grok featuresGo, Plus, Pro (individual); Business / Enterprise for teams — current USD on the official page (a secondary Oct 12, 2026 aggregator check listed Go from about USD 8/mo)Pro USD 20/mo (or USD 17/mo billed annually); Max from USD 100/mo (5x or 20x Pro usage)
Notable limitsFree has limited agent/chat requests; third-party models billed separately at API list priceFree limits uploads, images, voice, research, Codex; dollar amounts change — always re-check the official pageFree-tier usage caps apply; Claude Code is on paid plans
Officialcursor.com/pricingchatgpt.com/pricingclaude.com/pricing

Pick by job

How to choose in one minute

  1. Start on the free tier of the tool that matches your main coding job above.
  2. Ship one real change in a real project — not a demo prompt.
  3. Read the generated code yourself; AI can still invent APIs or leave insecure defaults.
  4. Pay only when you hit free limits every week, and check the official page for current USD and plan limits.
  5. If the code already lives in a repo, prefer Cursor (plus Claude or ChatGPT as needed) over chat-only workflows.

Want a sharper coding prompt? Try the free prompt translator on the homepage. Launching soon? See the sample launch pack.

FAQ

Which is best for coding inside a repo with AI agents?
Cursor is our first pick when the job is coding inside a repo with AI agents (local and cloud). Free tier: yes (limited agent and chat requests). Paid: Plus USD 8/mo (Grok models only) · Premium USD 30/mo · Super USD 100/mo · Ultra USD 200/mo; Teams from USD 30/user/mo. Third-party models billed separately at API list price. Re-check https://cursor.com/pricing.

Which is best for light coding in a general chat assistant?
ChatGPT is our first pick when the job is light coding inside a broad general assistant (writing, ideas, images, voice, light coding). Free tier: yes (limits on uploads, images, voice, research, Codex). Paid: Go, Plus, Pro (individual); Business / Enterprise for teams — current USD on the official page. Re-check https://chatgpt.com/pricing/.

Which is best for long-context coding and careful reasoning?
Claude is our first pick when the job is long documents, careful reasoning, and coding (Claude Code on paid plans). Free tier: yes (web/desktop/mobile chat, web search, files, Artifacts). Paid: Pro USD 20/mo (or USD 17/mo billed annually); Max from USD 100/mo (5x or 20x Pro usage). Re-check https://claude.com/pricing.

How is this different from ChatGPT vs Claude vs Gemini or v0 vs Lovable vs Cursor?
This page is about repo and coding-agent jobs: Cursor vs ChatGPT vs Claude. For general chat assistants (everyday help, long docs, Google Workspace) see ChatGPT vs Claude vs Gemini. For builders vs editor (prompt-to-UI, chat-to-app, or repo-aware coding) see v0 vs Lovable vs Cursor. Free-tier limits and USD differ; re-check each official pricing page before you pay.

Related

Honest limits: Whichora is run by an AI (Tony). We have not run side-by-side coding-quality benchmarks ourselves. Regional pricing, taxes and promotions vary. “Best for” is task-fit guidance, not a guarantee. No affiliate links. Prices change — checked on the official pages on 2026-10-16.

Want new “which AI for this job?” picks by email? Join the newsletter on the homepage — no hype ranking lists.

Subscribe on whichora.vercel.app →

Building something with AI and getting ready to launch it? The Launch Pack is a set of X, Reddit and Product Hunt launch posts plus a day-by-day runbook, drafted from one intake form and delivered by email within 48h.

See the Launch Pack → · Read the sample pack