ChatGPT vs Claude
Pick Claude if you mainly write, code or run long, multi-step work. Pick ChatGPT if you want one app that does everything, including images, video-call voice and a stronger free plan. Claude's newest model, Claude Opus 5.5, tops the independent Artificial Analysis Intelligence Index (58) and Claude models hold the top spot on the LMArena text leaderboard. ChatGPT has the wider feature set, an $8 Go plan and 900 million weekly users. Both main paid plans cost $20 a month.
At a glance
| ChatGPT | Claude | |
|---|---|---|
| Maker | OpenAI | Anthropic |
| Top model in the app | GPT-6 Astra (Pro reasoning, Pro plan) | Claude Fable 5.1 and Opus 5.5 (paid plans) |
| Free plan model | GPT-5.6 Luna, unlimited text chats | Sonnet and Haiku models, 5-hour usage limits |
| Paid plans | Go $8, Plus $20, Pro from $100 | Pro $20 ($17 yearly), Max $100 or $200 |
| Team plan | Business (price via sales) | Team $25/seat ($20 yearly) |
| Image generation | Yes, all plans (limited on Free) | No built-in image generator |
| Coding agent | Codex (all plans, different limits) | Claude Code (Pro and above) |
| Ads | Free US users since Jan 2026; Go may show ads | None |
| App Store rating (US) | 4.8/5 (about 11M ratings) | 4.7/5 (about 268K ratings) |
Head to head
Model intelligence Winner: Claude
On the Artificial Analysis Intelligence Index, which combines ten independent tests, Claude Opus 5.5 at max effort scores 58, the highest of any model it tracks. Claude Fable 5.1 and OpenAI's GPT-6 Astra tie on 53, and GPT-6 Sol scores 48. Opus 5.5 is included in Claude Pro at $20. ChatGPT's best model, Astra, needs the Pro plan from $100 for full 'Pro reasoning'. OpenAI's own launch figures for Astra on maths are very high (97.6% on FrontierMath Tier 4), but that is a vendor claim.
Coding and agents Winner: Claude
On Terminal-Bench 4.0, a test of coding tasks in a real command line, Anthropic reports 66.4% for Opus 5.5 against 57.9% for GPT-6 Astra. On AutomationBench the two are close (Opus 5.5 40.0%, Astra 41.4%). Claude Code is widely used by developers and is included from the $20 Pro plan. ChatGPT's Codex is strong too and comes with every plan, including Free, so it is the easier way to try an AI coding agent at no cost.
Writing and tone Winner: Claude
Claude models hold the top of the LMArena text leaderboard, where people vote blind on which answer they prefer. On 13 September 2026 Claude Fable 5 ranked #1 (1506) and Anthropic had seven of the top 15 places; the best GPT model then listed was GPT-5.6 Sol at #18 (1483). GPT-6 had not yet been ranked. Blind votes reward clear, well-structured answers, which is where Claude has built its reputation.
Features and multimedia Winner: ChatGPT
ChatGPT does more things in one app: image creation, voice chat with live video, deep research, memory, projects, scheduled tasks, agent mode, the Codex coding agent and ChatGPT Work for slides, documents and spreadsheets. OpenAI also runs the Atlas browser. Claude has no image or video generator. It does make finished Word, Excel and PowerPoint files, interactive artifacts and scheduled or parallel tasks, and it can use a browser or (in beta) your computer.
Free plan Winner: ChatGPT
ChatGPT Free gives unlimited text chats with GPT-5.6 Luna plus limited images, voice, deep research and Codex. The catch is ads for US users. Claude Free is ad-free and includes web search, file creation, memory and connectors, but it uses the smaller Sonnet and Haiku models, has no Claude Code and runs out faster because usage resets every five hours.
Paid plans and value Winner: ChatGPT
Both main plans cost $20 a month. ChatGPT has a cheaper first step (Go at $8) and a heavy-use Pro tier from $100 with 5x Plus usage (the $200 Pro tier has been closed to new sign-ups since 10 September 2026). Claude has no budget tier but offers Pro at $17/month billed yearly and two Max tiers ($100 for 5x, $200 for 20x Pro usage). For light users ChatGPT's ladder is easier. For developers, Claude Pro's included Claude Code and top models are very good value, so this round is close.
Research and web search Tie
Both apps search the web with citations and both run multi-step research that writes a cited report (ChatGPT deep research, Claude Research). ChatGPT gives Free users a limited taste of deep research; Claude Research needs Pro. Neither company publishes a head-to-head research benchmark, so we call this even.
Privacy and ads Winner: Claude
ChatGPT began showing ads to free US users on 17 January 2026 and says the Go plan may show them too. Claude shows no ads on any plan. Both let you control whether chats train future models, and both offer business plans that exclude company data from training. Anthropic also publishes the system prompts used in the Claude apps.
Which should you choose?
Choose ChatGPT if…
- You want images, voice with video and research in one app
- You want the most useful free plan, or a cheap $8 step up
- You try many different tasks and like having every tool in one place
- You want to try a coding agent (Codex) without paying
Choose Claude if…
- You write long documents, reports or fiction and care about tone
- You are a developer who will use Claude Code every day
- You want the top-scoring model (Opus 5.5) on a $20 plan
- You do not want ads in your AI app
This is the closest head-to-head in consumer AI, and the right answer depends on what you do.
- Claude wins on raw quality. Opus 5.5 leads the Artificial Analysis Intelligence Index, Claude tops LMArena's blind text votes, and it leads on the agentic coding tests both companies publish. If your work is mostly writing, coding or analysis, Claude Pro at $20 gives you more model for the money.
- ChatGPT wins on breadth. Images, voice with video, a generous free plan, an $8 Go tier, Codex on every plan and the Atlas browser. If you want one app for the whole family or for lots of small tasks, ChatGPT is the safer pick.
Who should not pick Claude: people who need image generation or the most generous free tier. Who should not pick ChatGPT: people who dislike ads or model changes every few months. Many heavy users pay for both.
Plans and prices side by side
| Tier | ChatGPT | Claude |
|---|---|---|
| Free | $0, GPT-5.6 Luna, ads in US | $0, Sonnet and Haiku, no ads |
| Budget | Go, $8/month | None |
| Standard | Plus, $20/month | Pro, $20/month ($17 billed yearly) |
| Heavy use | Pro, from $100/month | Max, $100 (5x) or $200 (20x) |
| Teams | Business, via sales | Team, $25/seat ($20 yearly); premium seat $125 |
| Enterprise | Via sales | $20/seat plus usage at API rates |
Prices are US web prices as of 23 September 2026. Apple in-app prices can be higher.
Benchmarks at a glance
| Test | Claude | ChatGPT (OpenAI models) | Source |
|---|---|---|---|
| Artificial Analysis Intelligence Index | Opus 5.5: 58; Fable 5.1: 53 | GPT-6 Astra: 53; GPT-6 Sol: 48 | Artificial Analysis |
| Terminal-Bench 4.0 (agentic coding) | Opus 5.5: 66.4% | GPT-6 Astra: 57.9% | Anthropic / OpenAI |
| AutomationBench | Opus 5.5: 40.0% | GPT-6 Astra: 41.4% | Anthropic |
| LMArena text (13 Sep 2026) | Fable 5: #1 (1506) | GPT-5.6 Sol: #18 (1483) | LMArena |
Vendor numbers are claims made at launch. Independent scores can move as new versions ship; GPT-6 had not yet appeared on LMArena when we checked.
How the apps feel day to day
ChatGPT picks a model for you and routes harder questions to reasoning models, so most people never touch a model menu. Claude lets paid users choose between Fable 5.1, Opus 5.5, Sonnet 5 and Haiku 4.5, and set an effort level.
Usage limits work differently. ChatGPT does not fully publish its caps. Claude gives each plan an allowance that resets every five hours; long chats, big files and stronger models use it up faster. Anthropic raised these limits for paid plans when Opus 5.5 launched on 22 September 2026.
- On Claude Pro, keep long document work in a Project so you do not re-upload files; re-sending big files burns through the five-hour allowance.
- On ChatGPT Free, turn off 'Improve the model for everyone' in Settings > Data Controls if you do not want your chats used for training.
- If you code, try both agents on the same small task: Codex is free to try in ChatGPT, while Claude Code needs Claude Pro ($20).
- Lower Claude's effort level for quick questions; higher effort is slower and uses more of your allowance.
- If you need images and strong writing, a common setup is Claude Pro for drafting plus ChatGPT Free for images.
Jargon explained
- Context window
- How much text an AI can keep in view at once, measured in tokens. A token is roughly three-quarters of a word.
- Artificial Analysis Intelligence Index
- An independent score that combines ten tests of reasoning, knowledge, maths and coding into one number, so different AI models can be compared fairly.
- LMArena
- A public site where people compare two anonymous AI answers and vote for the better one. The votes produce a ranking score.
- Terminal-Bench
- A test where an AI must complete real software tasks by typing commands into a computer terminal.
- Coding agent
- An AI tool that can read your code, edit files and run commands on its own to finish a programming task.
Frequently asked questions
Is Claude smarter than ChatGPT?
On current independent tests, yes, narrowly. Claude Opus 5.5 scores 58 on the Artificial Analysis Intelligence Index against 53 for GPT-6 Astra, OpenAI's top model. OpenAI's models still lead on some vendor-reported maths results.
Which is better for coding, ChatGPT or Claude?
Claude, for most developers. Opus 5.5 scores 66.4% on Terminal-Bench 4.0 against 57.9% for GPT-6 Astra, and Claude Code is included from the $20 Pro plan. ChatGPT's Codex is a good free way to start.
Can Claude make images like ChatGPT?
No. Claude can read images and make charts, diagrams and web pages, but it has no built-in image generator. ChatGPT creates images on every plan.
Which free plan is better?
ChatGPT's. It gives unlimited text chats with GPT-5.6 Luna plus limited images, voice and deep research. Claude's free plan has no ads but uses smaller models and lower limits.
Do they cost the same?
The main plans do: ChatGPT Plus and Claude Pro are both $20 a month. ChatGPT also has an $8 Go plan; Claude has no budget tier but offers $17/month if you pay yearly.
Does ChatGPT have ads?
Yes, for free users in the US since 17 January 2026, and OpenAI says the Go plan may include ads. Claude has no ads.
Sources
Every fact on this page comes from public information. Vendor figures are labelled as vendor claims.
- ChatGPT pricing (OpenAI)
- Claude pricing (Anthropic)
- Introducing Claude Opus 5.5 (Anthropic)
- GPT-6 Astra (OpenAI)
- Introducing GPT-6 Sol and Luna (OpenAI)
- LLM leaderboard: Intelligence Index (Artificial Analysis)
- Text Arena leaderboard (LMArena)
- ChatGPT on the App Store (Apple)
- Claude by Anthropic on the App Store (Apple)
- ChatGPT (Wikipedia)
- About ChatGPT Pro tiers (OpenAI Help Center)